Dynamic Magnetic Phenomena - Ar2003 PDF

Annual Report 2003
AG Magnetismus
TU Kaiserslautern
Address: Prof. Dr. Burkard Hillebrands
Fachbereich Physik
Technische Universität Kaiserslautern
Erwin-Schrödinger-Straße 56
67663 Kaiserslautern, Germany
Tel.: +49-(0)631-205-4228
Fax.:+49-(0)631-205-4095
Postal address: Postfach 3049

67653 Kaiserslautern, Germany
Internet: http://www.physik.uni-kl.de/hillebrands/
E-Mail: hilleb@physik.uni-kl.de
AG Magnetismus
TU Kaiserslautern
Our Group
From left to right:

Oskar Liedke, Bernd Pfaff, Sybille Müller, Alexander André,
Dr. Sergej O. Demokritov, Dr. Q.F. Xiao, Markus Weber,
Dr. Kurt Jung, Steffen Blomeier, Dr. Alexander Serga,
Dr. Vladislav Demidov, Dr. Galina Khlyap, Dr. Mikhail Kostylev (guest),
Andreas Beck, Dr. Mireia Blanco-Mantecón, Patricia Martin Pimentel,
Markus Laufenberg (guest), Christian Bayer, Dr. Jürgen Fassbender,
Hans Nembach, Dr. Björn Roos, Heike Schuster,
Dr. Thomas Wittkowski, Martin Rohmer, Dieter Weller,
Prof. Dr. Burkard Hillebrands
This report contains unpublished results and should

not be quoted without permission from the authors.
AG Magnetismus
TU Kaiserslautern
Contents
Contents
1 Introduction .......................................................................... 1
2 Personnel ............................................................................. 3
2.1 Members of the group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Visiting scientists, postdoctoral fellows and exchange students . . . . . . . . . . . . . . . 5
2.3 Guest seminars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.4 Visits of group members at other laboratories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.5 Group member photo gallery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3 Research Topics ..................................................................... 13
4 Equipment............................................................................ 17
5 Transfer of Technology............................................................. 21
6 Experimental Results ............................................................... 23
A. Dynamic Magnetic Phenomena .............................................. 23
6.1 Reflection and tunneling of spin waves through a region of inhomogeneous
magnetic field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6.2 Brillouin light scattering measurements of single micrometer sized mag-
netic elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
6.3 Observation of collective spin-wave modes in an inhomogeneously magne-
tized stripe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
B. Nonlinear Wave Effects ....................................................... 38
6.4 Experimental observation of symmetry-breaking nonlinear modes in an ac-
tive ring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6.5 Black soliton formation from phase-adjusted spin wave packets . . . . . . . . . . . . . . 43
6.6 Parametric generation of forward and phase-conjugated spin-wave bullets
in a magnetic film . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
C. Magnetic Films and Double Layers ......................................... 51
6.7 Surface smoothing and reduction of Néel “orange peel” coupling for mag-
netic tunnel junctions using low energy argon ions . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.8 High resolution magnetic patterning of exchange coupled multilayers . . . . . . . . 55
6.9 Measurement of the three magnetization vector components of a Ni81 Fe19
film . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
AG Magnetismus
TU Kaiserslautern
Contents
D. Exchange Bias Effect .......................................................... 62

6.10 Optical control of the magnetization in exchange biased NiFe/FeMn bilay-
ers on the picosecond timescale. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.11 Investigation of the polycrystalline Fe19 Ni81 /Fe50 Mn50 exchange bias sys-
66
tem with varying Cu spacer layer for partial decoupling . . . . . . . . . . . . . . . . . . . . . .
6.12 Structural analysis of the Ni81 Fe19 /Fe50 Mn50 exchange bias layer system . . . . 69
6.13 Modification of the magnetic properties of the polycrystalline
Ni81 Fe19 /Fe50 Mn50 exchange bias bilayer system by Ga+ -ion irradiation. . . . . 73
E. Elastic Properties............................................................... 77
6.14 Growth dynamics in sputtered BN films revealed by Brillouin light scatter-
77
ing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.15 Evidence of surface phonons in mesoscopic BN coated fibres . . . . . . . . . . . . . . . . 81
6.16 Monte Carlo method to analyze the dispersion of guided acoustic modes . . . . . 86
F. Transfer of Technology........................................................ 89
6.17 Mode-locked ps-Nd:YVO4 - laser for time-resolved measurements at wave-
lengths of 1064 nm, 532 nm and 266 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.18 The influence of amorphous carbon coating on encrustations of indwelling
catheter surfaces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
7 Publications .......................................................................... 95
8 Conferences, Workshops, Schools and Seminars ............................... 99
8.1 Conferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
8.2 Invited colloquia and seminars. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .100
8.3 Invited lectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .101
8.4 Contributions to other meetings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .101
Appendix: Impressions from our Group Ceremonies ..........................103
AG Magnetismus
TU Kaiserslautern
1 Introduction
Chapter 1: Introduction
Dear colleagues and friends,

again we report on our research from the last twelve months, the period of November 2002 to Oc-
tober 2003. Our university changed its name and is now a “Technische Universität” (University of
Technology) to better account for the focus on technology and natural sciences. As a consequence,
the university’s corporate identity was brought to an up-to-date standard which also affects our
Annual Report: you might notice that this year we have a new cover using the new university
layout.
We continue to work mostly in the field of magnetization dynamics and ion modification in mag-
netic systems, fields which generated a lot of interest in the last years. In recent years we have
moved from studies of eigen modes in small magnetic objects to dynamical properties in general,
and tried to establish the field of nonlinear spin waves in the general field of nonlinear wave phe-
nomena. A highlight is the discovery of symmetry breaking modes in an active ring structure,
which we were able to make after the development of phase sensitive parametric amplification. As
one example we studied so-called “Möbius” solitons, which need to travel twice about the ring to
meet the initial phase condition. We studied the process of spin wave reflection and transmission
from a region with a local field inhomogeneity and observed spin wave tunneling. In advancing
our experimental capabilities we have developed the technique of micro-Brillouin-light-scattering
allowing us now to investigate single micrometer sized magnetic elements. Progress was also made
in the field of ion irradiation of magnetic systems. We showed how modification of the coupling in
exchange bias system can be achieved by short heating pulses from a laser. In the Fe/Cr/Fe-trilayer
system we demonstrate the potential to magnetically pattern such structures on small length scales
using a focused ion beam machine.
Much of this work is embedded in the German priority program (“Schwerpunktprogramm”) enti-
tled “Ultrafast magnetization processes” funded by the Deutsche Forschungsgemeinschaft as well
as by the European Community Research Training Networks on “Ultrafast Magnetization Pro-
cesses in Advanced Devices” (ULTRASWITCH) and “New Exchange Bias Systems for Advanced
Magnetic Devices” (NEXBIAS). A third Research Training Network on “Ultrasmooth Magnetic
Layers for Advanced Devices”is currently in negotiation with the European Community and ad-
dresses new techniques for preparing ultra-smooth magnetic films for magnetic sensors. Applied
work was conducted in the European IST network “Low Power Magnetic Random Access Memory
with optimised Writing Time and Level of Integration” (NEXT). Since beginning of 2003 a new
research center “Materialien für Mikro- und Nanosysteme” has been established at our University
providing a basis for local funding.
Both Sergej Demokritov and Jürgen Fassbender received a call on new positions: Sergej received
a call on an associate professor position in the United States, which he had to refuse for sudden
personal reasons. Jürgen received an offer for a deputy research director position at the Institut für
Ionenstrahlphysik und Materialbearbeitung in Dresden. Congratulations to both!
Kurt Jung, who worked in our group since its beginning in 1995 on the position of an “Akademi-
scher Oberrat” and later of an “Akademischer Direktor” took early retirement and left the group
end of September. We wish him all the best for his new time.
AG Magnetismus
1
TU Kaiserslautern
1 Introduction
Two members of our group received their Ph.D.: Tim Mewes, who then started a post-doctoral stay
in the group of Chris Hammel in Ohio, and Stefan Poppe, who took a senior reseacher position at
the reseach centrum C.E.A.S.A.R. in Bonn.
Our work would not have been possible without valuable collaborations with people all over the
world. They are too many to list them here all. In particular we would like to thank, in alphabetical
order, Dima Berkow, Harry Bernas, Klaus Bewilogua, John Chapman, Claude Chappert, Darrell
Comins, Hajo Elmers, Claude Fermon, Jacques Ferré, Colm Flannery, Paulo Freitas, Loris Giovan-
nini, Volker Herzog, Boris Kalinikos, Gregor Kiefel, Mikhail Kostylev, Natalia Kreines, Norbert
Lampe, Wolfram Maaß, Jan Marien, Günter Marx, Roland Mattheis, Andrezj Maziewski, Jacques
Miltat, Alexandra Mougin, Fabrizio Nizzoli, Josep Nogues, Carl Patton, Günter Reiss, Günter
Reiße, Frank Richter, John R. Sandercock, Gerd Schönhense, Andreas Schreyer, Andrei Slavin,
Bob Stamps, André Thiaville, Stefan Visnovsky and Steffen Weißmantel for their interactions
with us and their strong input to our work. Collaborations within the Fachbereich Physik at the
University of Kaiserslautern (Martin Aeschlimann, Michael Fleischhauer, Henning Fouckhardt,
Wolfgang Hübner, Herbert Urbassek, Richard Wallenstein, Sandra Wolff, Christiane Ziegler and
their groups), the Institut für Oberflächen- und Schichtanalytik as well as the newly established
Nano+Bio-Center have been very stimulating. It was a pleasure to work together with Lumera
Lasers GmbH (Rolf Knappe) in setting up our new pulsed laser system. I am especially grateful
to Heinz Busch and Udo Grabowy and their start up company NTTF GmbH for the close contact.
I am much obliged to Stefan Sattel and his team from the TZO GmbH for providing convenient
general conditions for our work in Rheinbreitbach.
I would also like to thank all our sponsors, which are the Deutsche Forschungsgemeinschaft, the
Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie, the Humboldt Foun-
dation, the Deutscher Akademischer Austauschdienst, the European Community, INTAS, the State
of Rheinland-Pfalz and the University of Kaiserslautern. My special thanks go to Andreas Beck
and Sibylle Müller for their help in preparing this report.
It is my special pleasure to greet all former group members. May this report help to stay in touch
with each other.
If you are interested in our work I would be happy to hear from you. If you have any questions,
comments, suggestions, or any kind of criticism, please contact us.
With all my best wishes for Christmas, and a Happy New Year,
Kaiserslautern, November 2003
AG Magnetismus
2
TU Kaiserslautern
2 Personnel
Chapter 2: Personnel
2.1 Members of the group
Group leader:
Prof. Dr. Burkard Hillebrands
Senior scientists:
Dr. habil. Sergej Demokritov, Hochschuldozent
Dr. habil. Jürgen Fassbender, Privatdozent
Dr. Kurt Jung, Akad. Direktor until 09/03
Postdocs:
Dr. Mireia Blanco-Mantecón
Dr. Vladislav Demidov
Dr. Stefan Poppe until 08/03
Dr. Marc Rickart until 12/02
Dr. Alexander Serga
Dr. Q.F. Xiao since 07/03
Dr. Galina Khlyap since 10/03
Dr. Björn Roos since 08/03
Dr. Thomas Wittkowski
PhD students:
Dipl.-Phys. Christian Bayer since 02/03
Dipl.-Phys. Andreas Beck
Dipl.-Phys. Lisa Kleinen (Rheinbreitbach)
Dipl.-Phys. Patricia Martin Pimentel since 08/03
Dipl.-Phys. Tim Mewes until 11/02
Dipl.-Phys. Hans Nembach
Dipl.-Phys. Maciej Oskar Liedke since 11/02
Dipl.-Phys. Stefan Poppe until 05/03
Dipl.-Phys. Martin Rohmer since 04/03
Dipl.-Phys. Markus Weber since 11/02
Diploma Students:
Alexander André since 03/03
Steffen Blomeier since 12/02
Dirk Hoffmann since 09/03
Engineers and Technicians

Günther Moog (Rheinbreitbach)
Bernd Pfaff
Dipl.-Ing. (FH) Dieter Weller
AG Magnetismus
3
TU Kaiserslautern
2 Personnel
Secretaries:
Sibylle Müller
Heike Schuster (Forschungsschwerpunkt MINAS)
AG Magnetismus
4
TU Kaiserslautern
2 Personnel
2.2 Visiting scientists, postdoctoral fellows and exchange students

Prof. Boris Kalinikos, Electrotechnical University St. Petersburg,
21.11.02 - 22.11.02
St. Petersburg, Russia
12.05.03 - 13.05.03
Boris spent two short research stays in Kaiserslautern as a part of
our cooperation in the frame of an INTAS project. Discussions with
him have helped us a lot in understanding the origin of non-linear
excitations in active rings.
Prof. Robert L. Stamps, Department of Physics,

University of Western Australia, Perth, Australia 30.10.03 - 02.11.03
Bob spent a short research stay in Kaiserslautern. He helped a lot in
understanding ion irradiation induced modifications of the exchange
bias effect. He was also involved in the interpretation of magnetiza-
tion dynamics results.
Prof. J. Darrell Comins, Department of Physics,

University of the Witwatersrand, Johannesburg, South Africa 07.07.03 - 06.08.03
Darrell spent one month in Kaiserslautern, working successfully on
the interpretation of resonant surface excitations guided in layered
structures and their utilization for the determination of material pa-
rameters. Essential progress in the preparation of a comprehensive
review article on surface Brillouin scattering from bulk solids and
layered structures has been achieved.
Prof. Andrei Slavin, University of Rochester, Michigan, U.S.A. 04.06.03 - 22.06.03

Andrei spent a short research stay in our lab. He was heavily involved
in the modelling of localized modes of spin wave wells created in
confined magnetic objects.
Dimitry Kholin, Ph.D. student, Institute for Physical Problems,

Moscow, Russia 16.02.03 - 20.03.03
Dimitry is working on the fundamental properties of biquadratic and
non-collinear coupling in layered magnetic systems. He investigates
the interplay between the type of the coupling and the morphology
of interfaces.
Dr. Mikhail Kostylev, Electronics Faculty,

St. Petersburg Electrotechnical University St.Petersburg, Russia since 15.09.03
Mikhail is working on the problem of nonlinear spin waves fi-
nanced by a DFG project. In particular he works on theory of spin-
wave parametric amplification and generation in magnetic films and
magnetic-film-based active resonators by using pulsed microwave
parallel magnetic pumping.
AG Magnetismus
5
TU Kaiserslautern
2 Personnel
Markus Laufenberg, Ph.D. student,

INESC Microsistemas e Nanotecnologias, Lisboa, Portugal 06.10.03 - 24.10.03
Markus has prepared samples for our magnetooptical experiments
on the current induced transfer of magnetic momentum. During his
short stay in Kaiserslautern, which was a secondment within the EU
founded ULTASWITCH project, he was also involved in the Bril-
louin light scattering investigations of the prepared samples.
AG Magnetismus
6
TU Kaiserslautern
2 Personnel
2.3 Guest seminars

Prof. Theo Rasing Faculty of Natural Sciences, Math. and Informatics, University of
17.02.03 Nijmwegen, Niederlande
Ultrafast magnetization dynamics and switching
Physikalisches Kolloquium
Dr. Xiao Qunfeng Van der Waals-Zeeman Institute, Amsterdam, The Netherlands
12.05.03 Experimental study of nanoscale exchange coupling
Sonderkolloquium
Prof. Robert D. Shull NIST, Gaithersburg, MD, USA

13.06.03 Magnetic images of magnetization reversal in exchange-biased
bilayers
Sonderkolloquium
Dr. Claude Chappert CNRS, Université Paris-Sud, Orsay, France

16.06.03 Ultrafast switching of magnetic nanostructures: fundamentals and
applications
Physikalisches Kolloquium
Prof. J. Darrell Comins Department of Physics, University of the Witwatersrand,

21.07.03 Johannesburg, South Africa
High temperature light scattering in superionic fluorides
Sonderseminar
Dr. Radek Lopusnik NIST, Boulder, Colorado, USA

12.08.03 Magnetization dynamics study in ferromagnetic and semiconductor
materials
Sonderkolloquium
Dr. Britta Hausmanns Universität Duisburg-Essen

07.10.03 Magnetowiderstand und Ummagnetisierungsprozesse in einzelnen
nanostrukturierten magnetischen Leiterbahnen
Sonderkolloquium
Prof. Robert L. Stamps University of Western Australia, Australia

30.10.03 Five orders of dynamics: spin processes at 10−6 to 10−10 meters
Kolloquium des Schwerpunktes Materialien für Mikro- und
Nanosysteme
AG Magnetismus
7
TU Kaiserslautern
2 Personnel
2.4 Visits of group members at other laboratories

Steffen Blomeier University of Glasgow, United Kingdom
27.03.-27.04.09
Host: Prof. Dr. J. Chapman
Christian Bayer University of Minnesota, USA

02.04.-25.09.03
Host: Prof. P. Crowell
Steffen Blomeier University of Glasgow, United Kingdom

17.08.-14.09.09
Host: Prof. Dr. J. Chapman
Alexander André Oakland University, Rochester, USA

31.10.03-30.11.03
Host: Prof. Dr. A.N. Slavin
Sergej O. Demokritov University Perugia, Italy

06.03.03-20.03.03
Host: Dr. J. Gubbiotti
AG Magnetismus
8
TU Kaiserslautern
2 Personnel
2.5 Group member photo gallery
Alexander André Christian Bayer Andreas Beck

Diploma student Ph.D. student Ph.D. student
Dr. Mireia Blanco-Mantecón Steffen Blomeier Dr. Vladislav Demidov

Postdoc Diploma student Postdoc
Dr. habil. Sergej O. Demokritov Dr. habil. Jürgen Fassbender Prof. Dr. Burkard Hillebrands
Senior scientist and lecturer Senior scientist and lecturer Group leader
Dirk Hoffmann Dr. Kurt Jung Lisa Kleinen

Diploma student Senior scientist Ph.D. student
AG Magnetismus
9
TU Kaiserslautern
2 Personnel
Dr. Galina Khlyap Maciej Oskar Liedke Patricia Martin Pimentel

Postdoc Ph.D. student Ph.D. student
Dr. Tim Mewes Günther Moog Sibylle Müller

Postdoc Technician Secretary
Hans Nembach Bernd Pfaff Dr. Stefan Poppe

Ph.D. student Technician Postdoc
Dr. Marc Rickart Martin Rohmer Dr. Björn Roos

Postdoc Ph.D. student Postdoc
AG Magnetismus
10
TU Kaiserslautern
2 Personnel
Heike Schuster Dr. Alexander Serha Markus Weber

Secretary Senior scientist Ph.D. student
Dieter Weller Dr. Thomas Wittkowski Dr. Q.F. Xiao

Mechanical engineer Postdoc Postdoc
AG Magnetismus
11
TU Kaiserslautern
3 Research Topics
Chapter 3: Research Topics

The field of magnetism in films and multilayers is currently one of the strongest developing areas
in modern solid state physics. This is caused both by the challenging developments in the disco-
very and understanding of the basic physical phenomena, and by the strong impact into industrial
applications in the areas of sensors and information storage technology. New mechanisms like
interlayer exchange coupling, the giant magnetoresistance effect, the room-temperature tunneling
magnetoresistance effect, and, since very recently, spin current phenomena were discovered all
within the last one and a half decade. Applications based on these effects were developed, like
the magnetic read head based on the giant magnetoresistance effect found in nearly every hard
disk drive sold nowadays. The combination with microelectronics, the so-called field of magneto-
electronics is strongly expanding and bridging the gap between conventional magnetism and semi-
conductor physics in view of potential applications in sensor devices and magnetic random access
memories.
Most of our research projects are in this field. A main focus is on spin dynamics. We study the
eigen-frequency spectrum of excitations of the magnetization on the frequency scale using the
Brillouin light scattering technique, and the temporal evolution by time resolved magneto-optic
methods. We investigate high frequency properties like spin waves, time dependent magnetization
effects, and fast magnetic switching.
A key issue is the fabrication of high-quality epitaxial film and multilayer systems and devices
using molecular beam epitaxy as prototype systems to study fundamental problems.
In the field of applications we address problems of fast magnetic switching, the exchange bias
effect and tunneling magnetoresistance. We transfer our results into actual devices by working
closely together with industrial partners.
Magnetic films are very attractive and versatile nonlinear media. Considering spin waves in films
as one example of nonlinear waves we study nonlinear effects which are of a great importance for
nonlinear science in general.
As a second working area we develop and investigate carbon films for medical applications in the
framework of the Institute for Thin Film Technology in Rheinbreitbach. A special focus is on the
determination of elastic properties of hard coating materials.
Overview on projects
1) Epitaxial magnetic films and multilayers: growth, structure and magnetic properties
The preparation of samples with highest possible structural quality and characterization is very
important to be able to study magnetic phenomena with the necessary precision. We achieve
this by using molecular beam epitaxy employing the standard in-situ methods for chemical and
structural analysis. They comprise Auger spectroscopy for chemical analysis, low and high
energy electron diffraction, and in-situ scanning tunneling and atomic force microscopy. To
characterize the magnetic properties we perform in-situ Brillouin light scattering spectroscopy
and magneto-optic Kerr effect magnetometry. Ex-situ, the samples are investigated using Bril-
louin light scattering, vector Kerr magnetometry, vibrating sample magnetometry, and more.
Scientific subjects are magnetic anisotropies induced at interfaces and by controlled defects,
and interlayer coupling effects between magnetic films in multilayers. Special attention is paid
AG Magnetismus
13
TU Kaiserslautern
3 Research Topics
to the interplay between the morphology at the interfaces (atomic defects, steps, roughness and
interdiffusion) and the magnetic properties.
2) Surface smoothing
It is very important to fabricate films and multilayers with maximum degree of smoothness.
Undesirable roughness, for example, results in a reduced figure of merit in magnetoelectronic
devices. We develop a technology to smooth surfaces of the films after their preparation. For
this purpose we use low-energy beams of argon ions. Mono-energetic, low-energy ions allow
for a very controllable smoothing process of the surface without creation of an essential number
of defects.
3) Dynamic magnetic properties of laterally patterned nanostructures

We investigate the basic magnetic properties of systems patterned on the micrometer to nanome-
ter scale. In particular we focus on the domain structure and the change in the spin wave mode
spectrum due to lateral confinement effects. We have developed a Brillouin light scattering
setup, operating in a Fourier microscope like mode, to obtain sub-micrometer scale spatial in-
formation about the distribution of dynamic excitations in small magnetic objects. We also
developed a micro-focus Brillouin light scattering system to investigate single magnetic ele-
ments. Using those methods we have observed lateral quantization of spin waves in magnetic
stripes and rectangular elements. Main results are the observation of quantized modes and
of edge modes existing in areas with a large internal field gradient, and static and dynamic
coupling effects between magnetic objects. The experiments are accompanied by numerical
simulations.
4) Nonlinear properties of high-amplitude spin waves

Spin waves with high precession angles are an interesting object for the investigation of general
effects of nonlinear wave propagation in dispersive, anisotropic, and dissipative media. Con-
trary to nonlinear optical pulses, the spectrum of spin waves can be easily manipulated, by, e.g.,
changing the orientation and the value of the applied magnetic field. In addition spin waves are
much slower than light pulses making their observation easier.
Using the time-resolved Brillouin light scattering technique developed in our lab, we measure
the intensity distribution of spin waves propagating in a magnetic film with spatial and temporal
resolution. Central problems are: the amplification of spin waves in the linear and nonlinear in-
tensity regimes, the formation of instabilities (e.g. self-focusing), the propagation of nonlinear
excitations (solitons, magnetic “bullets”) and excitations in nonlinear media with a nontrivial
topology such as rings. An important development of these studies is the investigation of self-
generation of solitons and bullets in loops with an electronic feedback, the development of a
spin wave soliton “laser”, and the discovery of symmetry-breaking spin wave modes like the
“Möbius” solitons.
5) Fast magnetic switching

For memory devices it is of special importance how fast and secure magnetic domains can be
written and the magnetization of a single magnetic object can be reversed. The corresponding
time scale is in the picosecond to nanosecond regime. In order to investigate these pheno-
mena a time-resolved scanning magneto-optic Kerr microscope has been constructed. The time
evolution of the magnetization is sensed stroboscopically. The magnetization dynamics, spin
AG Magnetismus
14
TU Kaiserslautern
3 Research Topics
wave propagation effects and in particular the switching behavior of thin magnetic films and
nanostructures are investigated.
6) Magnetic nanopatterning
Light ion irradiation is an excellent tool to locally modify magnetic properties on the sub-
micrometer scale, without affecting the surface topography. This effect is used to magnetically
pattern ultrathin films and multilayers using resist masks patterned by electron beam lithogra-
phy. The major difference between this technique and conventional lithographic techniques is
that the environment of the nanostructures can also be magnetic (paramagnetic, antiferromag-
netic). A focus is on coupled magnetic systems, such as exchange bias bilayers and exchange
coupled trilayers. Using a focused ion beam we were able to pattern epitaxial Fe/Cr/Fe(001)
trilayers changing the the sign of the interlayer coupling with a lateral resolution better than
200 nm.
7) Exchange bias systems

The investigation of exchange bias systems is of fundamental as well as technological impor-
tance. The effect is a shift of the hysteresis loop along the field axis, and it appears in multilayers
of coupled ferromagnetic and antiferromagnetic films. We study in particular structurally well
characterized epitaxial bilayers. The role of defects and interfacial mixing is investigated using
ion irradiation in order to artificially create disorder. Ion irradiation techniques are also applied
to modify the magnitude and direction of the exchange bias field. TEM studies are carried out
to investigate the structural and magnetic properties as well as their dependency on the irradi-
ation with He+ and Ga+ ions. A picosecond all-optical pump-probe setup was developed to
study thermal activated unpinning of the exchange coupling at the FM/AFM interface. This is
of high technological interest, especially for magnetic sensor and storage applications.
8) Elastic properties of hard, super-hard and inhomogeneous films and multilayers

We prepare hard and super-hard films and investigate their elastic properties using Brillouin
light scattering. Research subjects are amorphous carbon (a-C:H and ta-C:H) and boron nitride
films, which are prepared using unbalanced magnetron sputtering. The elastic constants are de-
termined from the dispersion curves of surface and film phonons (Rayleigh and Sezawa modes).
One of our aims is to prepare hard and super-hard films with minimized internal stresses.
9) Biofunctionalized surfaces for medical applications

Amorphous thin carbon films are known to be very biocompatible, and they can be prepared
by various deposition techniques to qualify for miscellaneous applications in the biological and
medical field. At the Institute for Thin Film Technology we develop in close collaboration with
our spin off company NTTF GmbH biocompatible and biofunctionalized surfaces for medical
implants, surgical instruments and cellbiological equipment. Currently we are working on
carbon coatings for endwelling catheters and cell culture dishes (both made of temperature
sensitive polymers) as well as on the development of diffusion barrier coatings on polymers.
AG Magnetismus
15
TU Kaiserslautern
4 Equipment
Chapter 4: Equipment
A) Preparation and characterization of thin films and multilayers
1. multi–chamber molecular beam epitaxy system (Pink GmbH) comprising
a. deposition chamber
(electron beam and Knudsen sources, RHEED, LEED, Auger)
b. scanning tunneling and atomic force microscopy chamber
(in-situ STM/AFM, Park Scientific)
c. Brillouin light scattering and Kerr magnetometry chamber
(magnetic field 1.2 T, temperature range 80 − 400 K)
d. load lock chamber
e. preparation chamber
(optical coating, heating station 2300◦ C)
f. transfer chamber
g. atom beam oxidization chamber with in-situ four-probe resistively measurement
stage
2. two-chamber UHV multideposition system
a. deposition chamber
(electron beam and Knudsen sources, LEED, Auger)
b. ion beam sputtering chamber with ion beam oxidation module and mask system
3. two-chamber UHV deposition system with in-situ 5 keV ion beam irradiation station
4. two-magnetron sputtering system for hard coatings
5. atomic force microscope (Solver, NT-MDT)
6. clean room facility with flow box, spin coater, etc.
B) Magnetic characterization
1. vibrating sample magnetometer with alternating gradient magnetometer option
(magnetic field 1.6 T, room temperature)
2. vibrating sample magnetometer
(magnetic field 5 T, temperature range 2 − 350 K)
3. two vector Kerr magnetometers
(longitudinal and transverse Kerr effect, magnetic field 1.2 T, temperature range 2 − 350 K,
automated sample positioning)
4. high-field polar Kerr magnetometer
(magnetic field 5 T, temperature range 2 − 350 K)
5. time-resolved vector Kerr magnetometer (10 ps time resolution and microwave setup for
generation of short field pulses)
6. scanning Kerr microscope with time resolution
7. picosecond all-optical pump-probe setup (adjustable delay up to 6 ns; ps-laser Lumera
Lasers GmbH)
AG Magnetismus
17
TU Kaiserslautern
4 Equipment
8. magnetic force microscope with magnet (NT-MDT)

9. two Brillouin light scattering spectrometers, computer controlled and fully automated
(magnetic field 2.2 T) with stages for
a. low temperature measurements (2 − 350 K)
b. space-time resolved measurements for spin wave intensity mapping (resolution
50 µm, 0.83 ns)
c. micro-focus measurements (focus diameter 0.5 µm)
d. in-situ measurements
e. elastic measurements
10. microwave setup (up to 32 GHz) comprising a network analyzer, microwave amplifiers,
modulators, pulse generators, etc.
11. magnetotransport setups (magnetic field 1.5 T, temperature range 20 − 400 K)
C) Equipment at the Institute for Thin Film Technologies (IDST), Rheinbreitach
1. Preparation of thin films:

a. chemical vapor deposition (CVD) facility
b. physical vapor deposition (PVD) facility
c. plasma enhanced CVD (PECVD) facilities with an inductively coupled rf-plasma
beam source and several magnetrons of different sizes
2. Surface and thin film analysis:
a. profilometer: measurement of coating thickness and roughness determination of
intrinsic stress and Young modulus
b. Ball on Disk: measurement of friction coefficient analysis of surface friction
c. Revetest: determination of adhesive strength analysis of microcracks
d. microindention: determination of plastic and elastic microhardness (Vickers)
e. optical contact angle measurement: determination of solid surface free energy and
surface tension evaluation of hydrophobicity and hydrophilicity
f. reflection- and transmission-spectroscopy (UV-VIS): optical measurements with
wavelength range from 185 nm to 915 nm (resolution 1 nm), determination of ab-
sorption coefficient and optical gap (Tauc)
g. (environmental) scanning electron microscopy (ESEM)1 : comprehensive struc-
tural microanalysis of conducting, isolating, anorganic, organic and wet samples
h. energy dispersive X-ray microanalysis (EDX)1 : non-destructive fast analysis of
elements
i. neutron activation analysis (NAA)2 : qualitative und quantitative analysis of main
and trace components
j. elastic recoil detection analysis (ERDA)2 : analysis of trace elements with depth
resolution analysis of hydrogen content
1 in cooperation with NTTF GmbH, Rheinbreitbach
2 (accelarator
enhanced analysis in cooperation with the accelarator laboratories of the Universities of Munich,
Bonn and Cologne
AG Magnetismus
18
TU Kaiserslautern
4 Equipment
k. Rutherford Backscattering (RBS)2 : analysis of trace elements with depth resolu-

tion
l. synchrotron-X-ray-fluorescence (SYXRF)2 : non-destructive analysis of elements
AG Magnetismus
19
TU Kaiserslautern
5 Transfer of Technology
Chapter 5: Transfer of Technology
1. Magnetism
With our facilities within the Department of Physics at the University of Kaiserslautern we offer
consultance and transfer of technology in the areas of thin film magnetism, magnetic film structures
and devices, magnetic sensors, and in corresponding problems of metrology.
We are equipped to perform magnetic, transport, elastic and structural measurements of films and
multilayer systems.
This is in detail:
• magnetometry (magnetic field up to 5 T, temperature range 2 - 400 K) using vibrating sample
magnetometry, Kerr magnetometry, Brillouin light scattering spectroscopy
• magnetic anisotropies, optionally with high spatial resolution
• magneto-transport properties
• test of homogeneity of magnetic parameters
• exchange stiffness constants in magnetic films
• elastic constants
• surface topography
2. Institut für Dünnschichttechnologie (IDST) - Transferstelle der Tech-

nischen Universität Kaiserslautern, Rheinbreitbach
(Institute for Thin Film Technology - Center for Technology Transfer of
the University of Kaiserslautern, Rheinbreitbach)
As part of technology transfer the Institute of Thin Film Technology (IDST) offers among other
activities
• consultance in tribological problems
• development of product specific coatings
• optimization of coatings especially for medical applications
• coating of polymers and temperature sensitive materials
• coating of samples and small scale production series
• management for R&D-projects
The institute is located in Rheinbreitbach about 20 km south of Bonn in the Center for Surface
Technologies (TZO) to support the economy in the northern part of the Rheinland-Pfalz State.
AG Magnetismus
21
TU Kaiserslautern
5 Transfer of Technology
Address:
Institut für Dünnschichttechnologie
Maarweg 30-32
53619 Rheinbreitbach, Germany
Scientific director:
Prof. Dr. B. Hillebrands phone: +49 631 205 4228
e-mail: hilleb@physik.uni-kl.de
Contact:
Lisa Kleinen phone: +49 2224 900 693
fax: +49 2224 900 694
e-mail: kleinen@physik.uni-kl.de
Please contact us for more information.
AG Magnetismus
22
TU Kaiserslautern
6 Experimental Results
Chapter 6: Experimental Results
A. Dynamic Magnetic Phenomena

6.1 Reflection and tunneling of spin waves through a region of inhomoge-
neous magnetic field
S.O. Demokritov, A.A. Serga, A. Andre, V.E. Demidov, M.P. Kostylev, and B. Hillebrands1
Propagation of spin waves in an inhomogeneous magnetic field was discussed for the first time in
the sixties [1–4]. It was Schlömann, who first realized a close similarity between spin wave prop-
agation and a quantum mechanical particle [1]. In fact, neglecting the magnetic dipole interaction
and magnetic anisotropies, the Landau-Lifshitz equation describing magnetization dynamics can
be rewritten in the form of the Schrödinger equation with the dynamic magnetization m ∝ exp(iωt)
being the wave function and the magnetic field playing the role of potential energy:
2A ∂2 m ω
− + H(z)m = m , (1)
MS ∂z 2 γ
where A is the exchange stiffness, MS is the saturation magnetization, and γ is the gyromagnetic
ratio of the medium. For the simplest case of an unconfined magnetic medium the dispersion
spectrum of a plane spin wave (m ∝ exp(iqz)) can be then written as:
2h̄γA 2
hν = h̄ω = ∆(z) + q , (2)
MS
DC conductor
MW antenna
H0
Laser
beam
MW DC current
modulator source
MW Pulse
generator generator
Fig. 1: Schematic layout of the Brillouin light scattering setup in the forward scattering geometry with space
and time resolution for the study of spin wave propagation through an inhomogeneous magnetic field. For a
discussion of the components see the main text.
1 In collaboration with A.N. Slavin, Department of Physics, Oakland University, Rochester, Michigan, USA.
AG Magnetismus
23
TU Kaiserslautern
where ∆(z) = h̄γH(z) is the gap of the spectrum, which is reminiscent to the dispersion of a particle
in a potential field U(z),
h̄2 2
E = U(z) + q . (3)
2m
Thus, if a spin wave of frequency ω enters a region where the field H = H(z) varies, i.e., where
the gap ∆(z) is nonzero in Eq. (2), the wave keeps propagating through the inhomogeneous field,
albeit with changing wavevector, q = q(z) to fulfill the dispersion law Eq. (2). However, if the
value of the gap locally exceeds the value of the initial energy of the spin wave, there exists no real
wavevector anymore to fulfill the dispersion law for the given frequency ω. The wave is reflected
from this region, which thus can be considered as a potential barrier for the wave. Recently it
was shown that a strongly inhomogeneous internal field in magnetic micro-stripes resulting from
demagnetizing effects can cause turning points within the stripe which reflect spin waves and thus
create a spin wave well [5–7].
A quantitative analysis of the spin wave reflection from a field inhomogeneity taking into account
only the exchange interaction and neglecting the magnetic dipole interaction shows that the dy-
namic magnetization beyond the turning point is not zero: it just changes its dependence on z from
sinusoidal (m ∝ exp(iqz)) to exponential (m ∝ exp(−λz)) [4]. The spin waves tunnel through the
barrier.
Here we present a study of spin wave propagation in a magnetic film and the reflection from a
region with local field inhomogeneity. Contrary to previous studies [1, 2, 5] the magnetic dipole
interaction dominantly determines the properties of spin waves under consideration. We show that,
if the forbidden region is narrow enough, an essential spin wave tunneling process takes place. We
investigate the mechanism of the tunneling and demonstrate that it has a magnetic dipole origin.
The used experimental setup is schematically shown in Fig. 1. Spin wave packets in a transparent
yttrium ion garnet (YIG) film on a gallium gadolinium garnet (GGG) substrate are generated by
a microstrip antenna and are detected using the time- and space-resolved Brillouin light scatter-
ing (BLS) technique [8]. Both the homogeneous external field and the static magnetization are
oriented in the plane of the film parallel to the propagation direction of the spin waves, z. In this
case the dynamic magnetization components are mx and my . Such an orientation of the field and
the magnetization corresponds to the magneto-static backward volume wave (MSBVW) geome-
try, characterized by a negative group velocity of the spin waves [9]. The microwave excitation
part consists of a microwave generator and a modulator, which is controlled by a pulse generator
(pulse length 10 − 30 ns) and connected to the antenna situated on the YIG strip for spin wave gen-
eration. MSBVW spin-wave packets are generated with a carrier frequency ω = 2π · 7.095 GHz
and a carrier wavevector q = 210 − 220 cm−1 , the value of q being determined by the dispersion
relation of the spin wave in an external field of H0 = 1840 Oe. A narrow conductor of 50 µm diam-
eter mounted across the film carrying a dc-current is used to create a local inhomogeneous field,
H j (z). Depending on the direction of the dc-current, the total field (and, thus, the gap of the spin
wave spectrum) is locally either enhanced or reduced by the Oersted field of the current up to a
maximum field inhomogeneity of about ±200 Oe.
The idea of the experiment is further illustrated in Fig. 2. As mentioned above, the frequency of
the MSBVW spin wave decreases with increasing wavevector and the allowed states are situated
below the zero-wavevector gap ∆0 . Thus, to realize a scenario of spin wave reflection at a field
inhomogeneity, one should not increase, but decrease the field (gap). The inset of Fig. 2 illustrates
AG Magnetismus
24
TU Kaiserslautern
Fig. 2: Schematic spectrum

∆ (z) Hj of MSBVW spin waves in
H0 a magnetic film. ∆(z) is the
Energy
∆0 gap of the spectrum. The
w inset illustrates the profile
hω 0 of the gap, the locations z1
and z2 of the turning points
and the forbidden interval
∆0 z1 z2 z of width w due to the inho-
hω 0 mogeneous field caused by
the dc-conductor.
∆0 - ∆j
H0 + H j
H
0
H0 - H j
q
the geometry of the field, the profile of the gap, and the creation of the turning points z1 and z2 .
The turning points are determined be the condition h̄ω = ∆(H(z) = H0 − H j (z)). As it is seen
from the inset of Fig. 2, this equation has two solutions, z1 and z2 , and the interval between the
turning points is a forbidden region with no spin-wave state with ω. The width of the interval
w = z2 − z1 will be considered as the barrier width for spin wave tunneling. On the other hand,
an enhanced magnetic field does not essentially disturb the propagation: spin waves propagate
through the inhomogeneity by increasing the wavevector according to the local field.
Measurements of the propagation of spin-wave packets through the field inhomogeneity using the
space and time resolved Brillouin light scattering technique are displayed in Fig. 3. Two sequences
of snapshots are shown for different delay times. Figure 3a corresponds to an enhancement of the
local field by the dc-current, whereas the images shown in Fig. 3b were obtained at reduced local
field. Each snapshot displays the distribution of the spin-wave intensity over the film normalized
to its maximum.
Figure 3a demonstrates no essential reflection of the spin waves. As it is discussed above, a re-
gion with a slightly enhanced local field cannot contain a turning point for a MSBVW spin-wave
packet. The wave accommodates its wavevector according to the dispersion law and passes almost
unaffected through the region of field inhomogeneity. An observed weak reflection is due to the
non-adiabatical nature of the process: In fact, the above discussion implies that the wavelength
of the wave is much smaller than the lateral scale of the field inhomogeneity. In the experiment,
however, these two values are of the same order of magnitude.
A much more interesting effect is illustrated in Fig. 3b corresponding to a reduced local field.
One can easily see that the region of the reduced field inhibits propagation of spin waves. Due
to local field reduction the region of the inhomogeneous field serves as a forbidden region or
a barrier for the wave and reflects it. However, the spin-wave packet is partially reflected and
partially transmitted by the barrier. The transmission and reflection probabilities depend on the
barrier width and height and the wavevector of the wave. Indeed, this effect is reminiscent to the
quasi-classical quantum mechanical problem of particle reflection and tunneling.
To understand the physics of the observed spin wave tunneling the transmission coefficient, defined
AG Magnetismus
25
TU Kaiserslautern
Fig. 3: Propagation of a MS-

a) BVW spin-wave packet across a
YIG film with local field inho-
mogeneity, observed by space-
300
and time-resolved BLS. Snap-
250
shots of the spin-wave intensity
Field
0 200 at given delay times are shown.

1 150 a) field/gap has a local maxi-
(ns)
2
3 100 mum, creating a potential dip for
4
e
the wave; b) field/gap has a lo-
Tim
5 50
6 cal minimum, creating a poten-
7
z (m tial barrier for the wave. The
m) field profiles are displayed in the
b) bottom of each panel. The max-
imum absolute value of H j is the
same for the both panels and is
about 56 Oe.
300
250
0 200
Field
1 150
(ns)
2
3 100
4
e
Tim
5 50
6
7
z (m
m)
for the intensity of the wave, T , was measured as a function of the dc-current. In agreement with
the above consideration T is close to unity until the dc-current reaches a given value, jc , which
corresponds to the creation of a turning point just below the dc-conductor. Fig. 4a presents the
normalized barrier transmission coefficient, Tb = T ( j)/T ( jc ). As it is seen in the inset of Fig. 2,
by increasing the value of the dc-current the turning points can be moved away from each other.
The profile of the inhomogeneous field, the positions of the turning points and, accordingly, the
width of the barrier, w, can be easily calculated for a given j for the experimental geometry. For
the sake of clarity Tb is shown as a function of w. With increasing w the transmission coefficient
decreases. From a comparison of the measured dependency Tb (w) with the predictions of different
theoretical models the origin of the spin wave tunneling will be clarified.
Despite the obvious similarity with the quantum mechanic problem, the physics of the spin wave
tunneling is more complicated since the long-range magnetic dipole interaction important for mag-
netic systems must be taken into account. To begin with, let us first follow a simple approach
neglecting the dipole interaction [10]. In this case the Landau-Lifshitz equation for a sinusoidal
excitation m ∝ exp(iωt) can be rewritten in the form of the Schrödinger equation, Eq. (1). If H(z)
is a slow function of z, the solution of Eq. (1) can be written as as a wave m ∝ exp(iqz) with a
z-dependent wavevector obeying:
q2 (z) = (ω/γ − H(z))/D , (4)
where D = 2A/MS is the exchange stiffness constant (D = 5 · 10−9 Oe cm2 for YIG). Thus, the
solution of Eq. (1), m(z), is either a sinusoidal (H < ω/γ, outside the barrier) or an exponential
AG Magnetismus
26
TU Kaiserslautern
2.0
1.0
a) b)
0.8 1.5 w = 73 µm
Normalized magnetization, Dipole field

Barrier transmission coefficient
0.6
1.0
0.4
0.5
0.2
0.0 0.0
0 20 40 60 80 100 120 -0.1 0.0 0.1
Distance formthe barrier (cm)
Barrier width (µm)
Fig. 4: a) Barrier transmission coefficient as a function of the barrier width. Full circles: experimental values.
Dash line: results of the analytical approach, neglecting the magnetic dipole interaction. Solid line: results
of the numerical simulations. b) Profiles of the dynamic magnetization (full circles) and dipole field in the
wave. Incident, reflected, and tunneling waves are considered. The barrier is indicated by the hatched area.
The horizontal dotted line corresponds to the value of the magnetization amplitude without the barrier.
(H > ω/γ, inside the barrier) function of z. As it is seen from the above analysis, the possibil-
ity to penetrate into forbidden regions and tunnel through them is an intrinsic property of spin
waves even for the simplest model. However, this approach cannot explain the described exper-
imental observations. First of all, the predicted dependence of Tb (w) is exponential: Tb (w) =

exp(−2| z1z1+w q(z)dz)|). The dashed line in Fig. 4 is calculated based on the above expression.
It is clear that the prediction of this model disagrees with the experiment. Moreover, as follows
from Eqs. (1) and (4), the field must be increased to create a strongly reflective barrier. This is in
obvious contradiction with the experiment (see Fig. 3).
Next we consider a more realistic approach. It is known [11] that not the exchange, but the
magnetic dipole interaction dominantly determines the properties of spin waves with q ≈ 102 −
103 cm−1 in YIG films. Therefore, this interaction has to be taken into account in our analy-
sis. Unfortunately, due to the non-local origin of the interaction the analytical analysis is very
complex. Instead, a numerical approach has been used. The stationary case was considered and
the Landau-Lifshitz equation has been rewritten as a vector integral equation, which then has been
solved using the iterational convergence method. For excitations with a time dependence ∝ exp iωt
the equation for mx (z) yields:
∞

mx (z) = χ(z) × Gxx (z, z )mx (z )dz + hsx (z) (5)
−∞
where χ(z) = 4πMS H(z)/(H 2 (z) − (ω/γ)2 ) is the diagonal term of the complex magnetic suscep-
tibility of the medium taking into account the field non-uniformity. Gxx (z, z ) is a component of
the magnetostatic Green function in its real space presentation [12], and hs (z) is the magnetic mi-
crowave field of the antenna. Wave damping was taken into account in an usual way [11], i.e.,
by adding an imaginary part to the static magnetic field: H(z) → H(z) + i∆H, where ∆H is the
linewidth of ferromagnetic resonance of the film. In zero approximation the analytical solution of
Eq. (5) for spatially homogeneous static magnetic field H(z) = H0 is used:
AG Magnetismus
27
TU Kaiserslautern
∞
(χ0 − iχ0 ) exp(ikz)
hsx k
mx (z) = dk , (6)
−∞ [1 − χ0 (P(k) − 1)] + iχ0 (P(k) − 1)
where
hsx k is the spatial Fourier component of hs (z), χ0 = χ for H(z) = H0 , and P(k) is the
dipole matrix element calculated in [13]. A numerical algorithm providing fast convergence of the
iterational equation Eq. (5) with the start function Eq. (6) has been constructed. The Fast Fourier
Transform algorithm for discrete values of z and k was used followed by the Runge-Kutta technique
to solve the resulting system of differential equations. The numerically obtained dependence of
Tb (w) is shown in Fig. 4a by the solid line. The quantitative agreement between the theory and the
experiment is striking, especially if one takes into account that no fitting parameter was used in
the simulations. Thus, one can conclude that the observed effect is tunneling of dipolar spin waves
through the field inhomogeneity. It is important to mention that not only the dipole field, but the
dynamic magnetization as well tunnels through the barrier. This is illustrated by Fig. 4b, where
the calculated profiles of the dynamic magnetization and that of the dipole field are presented.
A standing wave on the left side of the barrier is due to interference between the incident and
reflected waves. It is in fact observed in the experiment if longer spin wave pulses, as shown in
Fig. 3, are used. But what is more important, the relation between the dynamic magnetization (full
circles) and the dipole field (dashed line) is almost the same outside as well as inside the barrier.
Thus, Fig. 4b clearly demonstrates that the observed tunneling effect is tunneling of the dynamic
magnetization.
Support by the Deutsche Forschungsgemeinschaft is gratefully acknowledged. The work is sup-
ported in part by the European Communities Human Potential programme under contract number
HRPN-CT-2002-00318 ULTRASWITCH.
References
[1] E. Schlömann, Spin-Wave Spectroscopy, in: Advances in Quantum Electronics, Ed. J.R. Singer (Columbia
University Press, New York, 1961).
[2] J.R. Eshbach, Phys. Rev. Lett. 8, 357 (1962).
[3] J.R. Eshbach, J. Appl. Phys. 34, 1298 (1963).
[4] E. Schlömann, J. Appl. Phys. 35, 159 (1964).
[5] J. Jorzick, S.O. Demokritov, B. Hillebrands, M. Bailleul, C. Fermon, K.Y. Guslienko, A.N. Slavin, D.V. Berkov,
N.L. Gorn, Phys. Rev. Lett. 88, 047204-1 (2002).
[6] J.P. Park, P. Earnes, D.M. Engebretson, J. Berezovsky, P.A. Crowell, Phys. Rev. Lett. 89, 277201-1 (2002).
[7] C. Bayer, S.O. Demokritov, B. Hillebrands, A.N. Slavin, Appl. Phys. Lett. 82, 607 2003.
[8] S.O. Demokritov, B. Hillebrands, A.N. Slavin, Phys. Rep. 348, 441 (2001).
[9] R.W. Damon, J.R. Eshbach, J. Phys. Chem. Solids 19, 308 (1961).
[10] E. Schlömann, R.I. Joseph, J. Appl. Phys. 35, 167 (1964).
[11] A.G. Gurevich, G.A. Melkov, Magnetization Oscillations and Waves (CRC Press, New York, 1996).
[12] K.Y. Guslienko, S.O. Demokritov, B. Hillebrands, A.N. Slavin, Phys. Rev. B 66, 132402 (2002).
[13] B.A. Kalinikos, A. Slavin, J. Phys. C.: Solid State Phys. 19, 7013 (1986).
AG Magnetismus
28
TU Kaiserslautern
6.2 Brillouin light scattering measurements of single micrometer sized mag-

netic elements
V.E. Demidov, S.O. Demokritov, and B. Hillebrands1
Brillouin light scattering (BLS) spectroscopy is well known as a powerful and flexible tool for the
investigation of dynamic magnetization processes, for a recent review see [1]. Among the main ad-
vantages of this technique the wide dynamic range and the potential to realize local measurements
of the amplitude of the dynamic magnetization are of importance. The BLS technique is especially
useful when investigations of dynamic magnetic properties of micrometer sized magnetic elements
are concerned. Recently, BLS spectroscopy has been successfully employed for studies of arrays
of micrometer sized magnetic wires and dots, see, e.g., [2, 3] and the references therein. However,
in a standard BLS experiment the diameter of the laser spot, which is used for probing the dynamic
magnetization, is much larger than the typical size of a magnetic element to be studied. As a result,
usual BLS measurements always demand large arrays of absolutely identical elements and cannot
be used for the investigation of a single one.
The purpose of this work is to elaborate a novel BLS setup with a probing laser spot in the submi-
crometer range, and to test this setup by measuring the dynamic magnetization in a single micro-
meter sized magnetic element.
The layout of the micro-BLS setup is shown in Fig. 1. An argon ion laser produces a vertically
polarized probing beam with a diameter of 2 mm. In order to reduce the divergence of the beam
the latter is expanded by a factor of ten by means of a beam expander and confined in diameter
by a diaphragm with an aperture of 3 mm. The resulting beam is transmitted through a thin-film
polarizer increasing the polarization degree of the beam up to 104 . Then the beam is reflected by
a polarizing beamsplitter and focused by a microscope objective on the surface of the magnetic
sample placed into a magnetic field applied parallel to its plane. The angle between the plane of
Argon ion Fig. 1: Layout of the micro-BLS setup.

Laser laser
light
Beam
White expander
light
CCD
Piezoelectric camera
stage
Diaphragm
Magnet
Thin-film
polarizer
Beamsplitters
To the
interfero-
meter
Microscope Polarizing
Sample objective beamsplitter
1 In collaboration with M. Laufenberg and P.P. Freitas, INESC Microsistemas e Nanotecnologias, Lisboa, Portu-
gal.
AG Magnetismus
29
TU Kaiserslautern
the sample and the axis of the objective is equal to 90◦ . The objective is characterized by the
magnification of 100×, the working distance of 4 mm, and the numerical aperture of 0.75. The
objective allows a focusing of the beam down to about 500 nm. In order to position the laser spot
to the desired place on the sample, the latter is mounted on a piezoelectric stage, which allows
positioning in two directions with a precision of about 50 nm. The total scanning range of the
stage amounts to a value of 100 µm. At the surface of the magnetic sample the light is inelastically
scattered by spin waves. As a result, the frequency of the scattered light is shifted by the value of
the frequency of the spin wave. In addition to that, the plane of polarization of the scattered light is
rotated by 90◦ [4]. The latter allows a partial separation of the weak scattered light from the very
intensive light reflected from the surface of the sample. This separation is fulfilled by the above-
mentioned polarizing beamsplitter, which transmits the horizontally polarized scattered light and
reflects the remaining part of the light back in the direction of the laser. The beamsplitter suppresses
the vertically polarized component in the transmitted light by a factor of 104 , which provides a
reasonable signal-to-noise ratio for the following analysis. Finally the scattered light is analyzed
by means of a tandem Fabry-Pérot interferometer operating in multipass configuration [5].
In order to find a micrometer sized magnetic element on the surface of a sample, which normally
has dimensions of a few millimeters, a viewing system is integrated into the BLS setup. This
system consists of a CCD camera supplied with a telescopic objective, two beamsplitters, and a
source of white light (see Fig. 1). It uses the reflected light separated from the scattered light by
the polarizing beamsplitter. The system allows the direct observation of the surface of the sample
and the position of the probing laser spot on the screen of a monitor during measurements.
The BLS technique is known for its selectivity with respect to the wavevector of a spin wave
[4]. For the case of scattering from non-confined media this selectivity is usually determined by
an aperture of the objective collecting the scattered light. For the case of a confined medium
such as a small magnetic element and/or the case of a finite size of the probing laser spot, the
selectivity is drastically affected by the uncertainty in the wavevector of the scattered light caused
by the spatial confinement. As far as in the micro-BLS setup the spot size w = 500 nm is very
small, the uncertainty in the wavevector of the scattered light is rather large and is of the value
of ∆k = 2π/w = 1.26 · 105 cm−1 . Taking into account the value of the numerical aperture of the
objective, NA = 0.75, and the wavelength of the incident light, λ = 514 nm, one can easily calculate
the largest wavenumber kmax of spin waves that can be detected by our setup. This value was found
to be equal to about 4.9 · 105 cm−1 . This indicates that keeping the axis of the objective at 90 ◦ to
the plane of the sample we simultaneously collect a signal from spin waves having wavenumbers
of up to 4.9 · 105 cm−1 . However, starting from a certain wavenumber kc < kmax the light scattered
by spin waves starts partially to leave the cone of rays passing the objective and the signal becomes
significantly smaller. For our case kc is equal to about 5 · 104 cm−1 .
The micro-BLS setup was tested using samples having a geometry, which is typical for integrated
magnetic memory circuits. Each sample contains 16 magnetic elements with dimensions of a · b =
1.3 · 2.3 µm2 . A schematic view of a magnetic element is shown in the inset to Fig. 2. The elements
consist of two magnetic layers, separated by a 4 nm thick Cu spacer. The bottom magnetic layer
is a 10 nm thick CoFe film, whereas the top one is 5 nm thick NiFe film. The elements are placed
at the top of a 40 µm wide and 60 nm thick Al stripe, which plays the role of the bottom electrode.
The 60 nm thick upper electrode has the shape of a narrow stripe crossing the magnetic element
along its short side. The width of the upper electrode is chosen to be 0.4 µm in order to leave a large
enough part of the upper magnetic layer accessible for the BLS measurements. During the BLS
measurements the laser spot was kept in the middle of the uncovered area of the upper magnetic
AG Magnetismus
30
TU Kaiserslautern
layer. The external magnetic field of strength H was applied along the long side of the magnetic
element.
Figure 2 shows a typical spectrum of thermal spin waves accumulated during 30 minutes (thick
solid line). As seen from Fig. 2, the signal from thermal waves is rather weak but, nevertheless,
it can be clearly seen above the level of noise. The small measured intensity of the scattered light
originates from the very small power of the probing laser light, which can be focused to 500 nm
diameter on the surface of the sample without danger to overheat the latter. For different samples
the maximum power level depends on the thermal conductivity of the substrate. For the samples
investigated here this value was found to be equal to 2 mW.
As seen from Fig. 2, despite the small dimensions of the magnetic element, the spectrum of ther-
mal waves does not demonstrate clear resonant peaks and appears as a continuous one. In order
to understand this fact the theoretical resonant spectrum was calculated for the element using the
dispersion law for spin waves propagating arbitrarily to the direction of magnetization in tangen-
tially magnetized ferromagnetic films [6]. The resonant frequencies were found neglecting the
non-uniformity of the internal magnetic field. The quantization of the in-plane wavevector k of
spin waves was calculated using the magnetic wall boundary conditions for dynamic magnetiza-
tion at the edges of the element: k2 = (nπ/a)2 + (mπ/b)2 , where n, m = 1, 2... The frequencies of
the 16 lowest-order resonances are shown in Fig. 2 by the vertical solid lines. These resonances are
divided into four sets s1-s4. The different sets are characterized by different values of n, whereas
the resonances within each set are characterized by different values of m. The resonances shown by
the strongest lines correspond to m = 1. With the increase of m the strength of the corresponding
lines decreases. The resonant spectrum was calculated for the strength of an internal magnetic field
Hi = 385 Oe, which was used as an adjustable parameter in order to fit the theoretical spectrum to
the experimental one. The indicated value of Hi corresponds to the frequency of the uniform fer-
romagnetic resonance in a tangentially magnetized ferromagnetic film fP equal to 5.66 GHz. This
frequency is indicated in Fig. 2 by a vertical dashed line.
The calculations show that the values of the wavenumbers corresponding to the shown resonant
frequencies lie in the range from 2.7 · 104 to 1.1 · 105 cm−1 . As far as these values are smaller than
kmax all the corresponding resonant peaks must be seen in the spectrum. However, every resonant
peak has a finite width, which is determined by the intrinsic magnetic losses in the magnetic
material. For NiFe this width is larger than 170 MHz. As seen from Fig. 2, this width significantly
exceeds the frequency separation between the peaks. As a result, the individual peaks merge one
180 Fig. 2: Measured spectrum of ther-

Magnetic element
mal spin waves in the magnetic ele-
160 Laser ment (thick solid line) and the calcu-
spot H
140 lated resonance spectrum for the same
Bottom Top element (thin vertical solid lines). In-
Intensity, a.u.
120 electrode electrode set: Schematic view of the single, mi-

crometer sized magnetic element.
100
s1 s2 s3 s4
80
60
fP
40
4 5 6 7
Frequency, GHz
AG Magnetismus
31
TU Kaiserslautern
with another and the spectrum turns out to be continuous.

It is important to note here that the thermal spectrum shown in Fig. 2 demonstrates approximately
the same intensity for the first two sets of resonances, whereas the third and the fourth sets have
significantly smaller intensities decreasing stepwise from one set to another. This fact confirms
the above-made conclusion concerning the value of kc = 5 · 104 cm−1 . The first and the second
sets of resonances are characterized by the wavenumber of the lowest-order resonance equal to
2.8 · 104 cm−1 and 5.0 · 104 cm−1 , respectively. For the third and the fourth sets these values are
7.4 · 104 cm−1 and 9.8 · 104 cm−1 . Consequently, all the light scattered by spin waves of the first
two sets is collected by the objective, whereas a significant part of the light scattered by spin waves
of the third and the fourth set is lost.
In conclusion, we have elaborated the micro-BLS setup allowing measurements of single micro-
meter sized magnetic elements. The size of the laser spot, which plays the role of a probe in the
BLS measurements, was reduced down to 500 nm. Despite the serious reduction of the probing
light power, a spectrum of thermal spin waves can easily be measured by means of the elaborated
setup in reasonable time. It is found that the reduction of the probing spot leads to the loss of
the resolution on wavenumber of spin waves. This fact limits one in the use of the micro-BLS
technique for measurements of resonant spectra of spin waves in small magnetic elements. Never-
theless, this disadvantage does not hamper the use of the micro-BLS technique for a local probing
of amplitudes of excited spin waves on the sub-micrometer scale like it was previously realized on
the sub-millimeter scale.
This work was supported in part by the priority programme SPP1133 “Ultrafast magnetization phe-
nomena” of the Deutsche Forschungsgemeinschaft, by the European Community under contract
IST-2001-37334 “Low Power Magnetic Random Access Memory with Optimised Writing Time”
(NEXT), and by the European Communities Human Potential programme under contract number
HRPN-CT-2002-00318 ULTRASWITCH. Young Researcher M. Laufenberg acknowledges the fi-
nancial support from the European Communities Human Potential Programme.
References
[1] S.O. Demokritov, B. Hillebrands, A.N. Slavin, Phys. Rep. 348, 441 (2001).
[2] S.O. Demokritov, B. Hillebrands, A.N. Slavin, IEEE Trans. Magn. 38, 2502 (2002).
[3] Y. Roussigne, S.M. Cherif, C. Dugautier, P. Moch, Phys. Rev. B 63, 134429/1 (2001).
[4] M.G. Cottam, D.J. Lockwood, Light scattering in magnetic solids, Wiley, New York (1986).
[5] B. Hillebrands, Rev. Sci. Instr. 70, 1589 (1999).
[6] M.J. Hurben, C.E. Patton, J Magn. Magn. Mater. 139, 263 (1995).
AG Magnetismus
32
TU Kaiserslautern
6.3 Observation of collective spin-wave modes in an inhomogeneously mag-

netized stripe
C. Bayer and B. Hillebrands1
Soft magnetic materials such as permalloy often have a very simple magnetic microstructure which
is not single domain. It is physically interesting to ask whether such systems have collective
excitations, i.e. spin-wave modes. We have studied the dynamic properties of inhomogeneously
magnetized 18 nm thick and 2.3 µm wide permalloy stripes using time-resolved Kerr microscopy
[1]. If the magnetic field, which is applied along the short axis of the stripe, is close to the shape
anisotropy field, we find a new class of modes at higher frequencies than that of the ordinary
precessional mode at the center of the stripe. Although these modes are detected at the edges of
the stripe, they are fundamentally different than the localized edge modes observed previously at
higher fields [2–4]. The localized modes are confined in regions of inhomogeneous magnetic field
at the edges of the stripe, and their eigenfrequencies are lower than the frequency of the center
mode. In contrast, we will show that the new modes, which we call “cross-over” modes, span the
entire stripe, but are, with the experimental setup used here, only detectable at the edges, where
the magnetization rotates by 90 degrees. The width of the edge regions determines the effective
wavevector near the edges of the stripe.
The detailed experimental procedure is described elsewhere [3, 5]. Here the applied field Ha was
oriented along the width of the stripe while the Gaussian field pulse Hpulse with an amplitude of
≈5 Oe and a temporal width of 120 − 150 psec was directed along the length of the stripe. Data
obtained in a magnetic field Ha = 80 Oe with the objective focused on the center of the stripe
are shown in Fig. 1(a). The response observed corresponds to a nearly uniform precession of the
magnetization about the effective field at the center of the stripe. As can be seen in the Fourier
transform of the data shown in Fig. 1(b), the spectrum in this case is dominated by a single peak. As
the observation point is moved away from the center, the spectrum changes as shown in Figs. 1(d)
and (f), with additional peaks appearing at higher frequencies. A full spectral image of a scan
Fig. 1: Time and frequency re-

sponse measured at different po-
sitions relative to the center of
the stripe.
1 The experimental part of his work was done at the School of Physics and Astronomy, University of Minnesota
in collaboration with J.P. Park, H. Wang, M. Yan, C.E. Campbell, and P.A. Crowell, School of Physics and
Astronomy, University of Minnesota, Minneapolis, Minnesota 55455, USA.
AG Magnetismus
33
TU Kaiserslautern
Fig. 2: Spectral images of cross section of a stripe in different applied fields. The insets show the spectra at the
position indicated by the dotted line. Note that the width of the images is larger than the physical width of the
stripe due to the diameter of the optical spot (≈ 800nm).
across the stripe can be constructed by measuring the spectrum at each position and converting the
spectral intensity to a gray-scale. The results for several fields from 25 Oe to 130 Oe are shown
in Fig. 2. The spectral intensity has been normalized by the integrated power measured at each
position. This procedure emphasizes the modes near the edges of the stripe, which are much
weaker than the center mode in absolute intensity. The spectra at low and high fields are well
understood. At 25 Oe, the magnetization is nearly parallel to the long axis of the stripe, and the
observed response is a single precessing mode. At high fields (> 130 Oe), the response consists
of a precessional mode spanning most of the stripe and a lower frequency mode localized at each
edge [2–4]. As discussed previously, the high-field mode structure can be understood in a model
in which the internal field is non-uniform but the magnetization is essentially uniform. Here we
focus on the field range of 60 − 130 Oe, in which clear modes are observed near the edges but at
higher frequencies than the center mode. This separate set of modes can be seen in Fig. 2 at 70 Oe
and 90 Oe. The insets of the top three panels of Fig. 2 show spectra obtained with the objective
positioned 1.2 µm from the center of the stripe, corresponding to the position indicated by the
dotted line.
All of the mode frequencies observed at different positions on the stripe are shown in Fig. 3(a) as a
function of the applied magnetic field. The frequencies shown are those of peaks with an amplitude
at least 5 times the background noise level. The solid curve is the frequency of the ordinary
FMR mode of the stripe, which was calculated under the assumption that the magnetization is
uniform and that it undergoes coherent rotation. The shape anisotropy energy was determined by
approximating the stripe as an ellipsoid with demagnetizing factors: Nx = Hd /Ms , Ny = 0, and
Nz = 4π − Hd /Ms , where the demagnetizing field Hd =60 Oe was taken from the location of the
experimental minimum in the FMR frequency. The points along the dashed curve at high fields are
the localized modes discussed above. The points at higher frequencies from 60 − 120 Oe are due
to the cross-over modes that appear near the edges of the stripe in Fig. 2.
The key observation about the field regime, in which the additional edge modes are observed
in Fig. 2, is that, unlike at higher fields (> 130 Oe), the magnetization of the stripe is strongly
nonuniform. The local magnetization direction is determined by a competition between the shape
AG Magnetismus
34
TU Kaiserslautern
Fig. 3: Measured frequencies as a

function of the applied field. The solid
Fequency (GHz) 4 curve is the frequency of the ordinary
FMR mode calculated as described in
the text. The points along the dashed
curve are the localized modes.
2
0
40 80 120 160
Applied Field (Oe)
anisotropy field, which favors a magnetization parallel to the edges of the stripe, and the applied
field. Fig. 4(a) shows the two components of the magnetization of a 2.3 µm wide stripe calculated
micromagnetically for an applied field of 90 Oe. The magnetization at the center is aligned parallel
to the applied field while it is nearly parallel to the stripe at the edges. Between these two extremes
the magnetization rotates smoothly over a length of 0.5 µm. Fig. 4(b) shows the corresponding
effective field He f f , including the applied, demagnetizing, and exchange fields. He f f is non-zero
only in the region where the magnetization is parallel to the applied field.
In addition to the experiment, simulations of the full Landau-Lifshitz-Gilbert equation were con-
ducted using the object-oriented micromagnetic framework (OOMMF) [6] using a realistic pulse
profile. The results of the simulations are compared with the experiment in Fig. 5 at fields of 90
and 125 Oe. Panels (a) and (c) show experimental spectral images. Panels (b) and (d) show spec-
tral images obtained from the time-domain simulations, which were processed in the same manner
as the experimental data. The simulations show some of the important qualitative features of the
experiment, but there are several discrepancies, the most important one is the apparent shift in field
scale. The mode structure near the edges of the stripe is qualitatively similar in both cases, with
the weight moving closer to the edges with higher mode index as noted above. The structure in
the center of the stripe seen in the simulations cannot be observed in the experiment. As the nodes
of the dynamic magnetization are closer together in the center of the stripe than at the edges the
Fig. 4: (a) The two components of the

static equilibrium magnetization are
shown as a function of the position x
in an applied field of 90 Oe. (b) The
corresponding effective field He f f . (c)
The wavevector, evaluated from the
dispersion relation as described in the
text, is shown as a function of position.
The inset shows the coordinate system.
AG Magnetismus
35
TU Kaiserslautern
Fig. 5: The upper panels (a) and (c) show

experimental spectral images of the cross-
section of the stripe at fields of 90 and
125 Oe. The lower panels (b) and (d) show
simulated images obtained as described in
the text. The spectra are normalized by the
integrated power at each position.
effective wavevector of the spin wave must be smaller there. As the diameter of the optical spot
(≈800 nm) used in the experiment is larger than the effective wave length in the center of the stripe
the Kerr signal averages out. For higher modes, the distance from the edge of the stripe to the
first internal node decreases, which is why the spectral weight observed in the experiment moves
further towards the edges of the stripe as the frequency increases.
A detailed picture of the features of spin wave in the region of inhomogeneous magnetization can
be obtained by considering the dispersion relation at each position in the stripe. This is the same
philosophy applied in the WKB argument used previously to described localized spin-waves [2,3],
except that in the current case both the internal field and the dipole-dipole matrix element vary
with position. The dynamic torque in our experiment couples only to wavevectors along the short
axis of the stripe. For a given orientation of M, the dispersion relation for spin-waves propagating
along the short axis can be calculated following Kalinikos and Slavin [7]. The dispersion relation
at three positions across the stripe in a field of 90 Oe is shown in Fig. 6. These positions are also
indicated by the labels a, b, and c in Fig. 4.
At the edge of the stripe, the magnetization is nearly perpendicular to the wavevector k, and the
corresponding dispersion relation, indicated (a) in Fig. 6, is that of the usual Damon-Eshbach
modes. In the center of the stripe, M||k, and the dispersion relation (c) is that of the magnetostatic
backward volume wave (MSBVW) modes [7]. The frequencies of the three lowest modes found in
simulations are shown as the dashed lines in Fig. 6. Near the edge of the stripe there is a minimum
wavevector determined by the intercept of the dashed curve with the Damon-Eshbach dispersion
relation (a). The wavevector determined as a function of position over the entire width of the
stripe is shown for the first cross-over mode in Fig. 4(c), in which the wavevector at the center of
the stripe is about 10 times larger than at the edge. The cross-over modes are thus characterized
by two different regimes: a dipolar region near the edge of the stripe and an exchange-dominated
regime of large wavevector near the center of the stripe. In contrast to the two higher modes, the
lowest frequency mode does not intercept the dispersion relation (c) at the center of the stripe.
This is a signature of the incipient spin-wave localization and becomes more pronounced as the
MSBVW dispersion relation at the center of the stripe shifts upward with increasing applied field.
Unlike the cross-over modes, the localized modes do not propagate in the center of the stripe. This
is the important distinction between the two types of modes, even though they are both observed
at the edges.
As in the spin-wave localization problem, it is not obvious a priori that the usual adiabatic condi-
tion for a WKB theory is satisfied [8]. In the edge regions, the dispersion relation changes on a scale
AG Magnetismus
36
TU Kaiserslautern
Fig. 6: The dispersion relation cal-

culated at three points along a cross-
section of the stripe of width 2.3 µm in
an applied field of 90 Oe. a, b, and c
indicate positions 1.0, 0.75, and 0 µm
from the center of the stripe. The three
dashed lines indicate the frequencies
of the lowest modes found in simula-
tions.
comparable to the wavelength, and the boundary conditions are not known exactly. Nonetheless,
for a given frequency it is straightforward to calculate the wavevector of each mode as a function
of position using the magnetization profile shown in Fig. 4. The Bohr-Sommerfeld integral
w/2
φ(n) = kn (x)dx, (1)
−w/2
can then be calculated, where w is the width of the stripe and n is the number of nodes. We
have calculated φ for the modes at 90 Oe with 2 - 16 nodes, finding φ(n) = φ0 + nπ, with φ0 =
(0.83 ± 0.02)π. The phase difference φ(n) − φ(n − 1) differs from π by at most 5%. The relative
success of the WKB approach for the cross-over modes suggests that it is meaningful to think of a
local dispersion relation, even in cases of a strongly inhomogeneous magnetization.
A full report has been submitted to Physical Review B [1].
C.B. acknowledges support by the University of Minnesota MRSEC NSF (DMR 0212032) during
his stay in Minneapolis and by the Studienstiftung des deutschen Volkes. This work was also
supported by NSF DMR-9983777.
References
[1] C. Bayer, J.P. Park, H. Wang, M. Yan, C.E. Campbell, P.A. Crowell, submitted to Phys. Rev. B.
[2] J. Jorzick, S.O. Demokritov, B. Hillebrands, D. Berkov, N.L. Gorn, K. Guslienko, A.N. Slavin,
Phys. Rev. Lett. 88, 047204 (2002).
[3] J.P. Park, P. Eames, D.M. Engebretson, J. Berezovsky, P.A. Crowell, Phys. Rev. Lett. 89, 277201 (2002).
[4] C. Bayer, S.O. Demokritov, B. Hillebrands, A.N. Slavin, Appl. Phys. Lett. 82, 607 (2003).
[5] J.P. Park, P. Eames, D.M. Engebretson, J. Berezovsky, P.A. Crowell, Phys. Rev. B 67, 020403R (2003).
[6] M.J. Donahue, D.G. Porter, OOMMF Users Guide, Version 1.0, in Interagency Report No. NISTIR 6376,
National Institute of Standards and Technology, Gaithersburg, Maryland, 1999.
[7] B.A. Kalinikos, A.N. Slavin, J. Phys. C 19, 7013 (1986).
[8] L.D. Landau, E.M. Lifshitz Quantum Mechanics: Non-Relativistic Theory, 3rd ed. (Pergamon, Oxford, 1977).
AG Magnetismus
37
TU Kaiserslautern
B. Nonlinear Wave Effects

6.4 Experimental observation of symmetry-breaking nonlinear modes in an
active ring
S.O. Demokritov, A.A. Serga, V.E. Demidov, and B. Hillebrands1
Solitons are large-amplitude, spatially confined wave packets in nonlinear media. They occur in
a wide range of physical systems, such as water surfaces, optical fibres, plasmas, Bose-Einstein
condensates and magnetically ordered media [1, 2]. A distinguishing feature of soliton behavior,
that is common to all systems, is that they propagate without a change in shape owing to the
stabilizing effect of the particular nonlinearity involved [1,3]. When the propagation path is closed,
modes consisting of one or several solitons may rotate around the ring, the topology of which
imposes additional constraints on their allowed frequencies and phases [4, 5]. Here we measure
the mode spectrum of spin-wave solitons in a nonlinear active ring constructed from a magnetic
ferrite film. Several unusual symmetry-breaking soliton-like modes are found, such as “Möbius”
solitons, which break the fundamental symmetry of 2π-periodicity in the phase change acquired
per loop: a Möbius soliton needs to travel twice around the ring to meet the initial phase condition.
Despite the numerous publications on nonlinear wave propagation [1–3, 6, 7], multi-soliton modes
in confined nonlinear systems have received very little attention to date. Very recently Carr et
al. have analyzed multi-soliton modes in a box with infinitely high walls and, equivalently, a
ring [4, 5], and they find that the symmetry of these modes can differ from that of linear modes of
the system. The linear modes of a box are sinusoidal standing waves enumerated by the number of
half-wavelengths that satisfy the box size. There is a direct connection between the mode number
of a linear mode and its mirror symmetry with respect to the center of the box: odd and even
modes are symmetric and antisymmetric, respectively. It was shown [5] for the case of a so-called
Fig. 1: Schematic layout of the ac-

tive ring based on an yttrium iron
Output
garnet film as the nonlinear medium
for spin-wave propagation and the ex-
perimental arrangement for observa-
Mixer
& A tion of nonlinear modes in the ring.
Detector The ring consists of the film waveg-
uide, two antennae, and the linear
CW microwave amplifier (A). Nonlinear
modes of the ring are filtered us-
ing parametric pumping at double fre-
Phase H quency at the position of the center an-
Shifter tenna as described in the text. The di-
rectional coupler, the phase-shifter and
Freq. 2 MW Pulse
the mixer with detector are used for
Doubler Switch Generator phase sensitive detection of the modes
MW by mixing them with a c.w. reference
Generator signal.
1 In collaboration with M.P. Kostylev and B.A. Kalinikos, St. Petersburg Electrotechnical University, 197376
St. Petersburg, Russia.
AG Magnetismus
38
TU Kaiserslautern
attractive nonlinearity necessary for solitons to form that nonlinear modes can be classified by the
number of solitons comprising a particular mode. However, there is no general connection between
the symmetry properties of a mode and the number of solitons. For example, two modes consisting
of two solitons are reported: one mode is antisymmetric and the other is symmetric with respect
to the box center. The symmetry of the former mode corresponds to the symmetry of the linear
mode with two half-wavelengths, while the latter mode has no linear counterpart and was therefore
termed a “symmetry-breaking mode”.
As a model system, we constructed an active nonlinear ring based on spin-wave propagation in
a magnetic low-loss film with longitudinally applied field. This system exhibits the attractive
nonlinearity necessary for soliton formation, and thus the nonlinear eigenmode spectrum consists
of sets of soliton-like wavepackets. (We use the term “soliton-like mode” because the excitations
found are not true solitons owing to the residual lossy character of the ring outside the location
of amplification. The difference between soliton-like modes and true solitons has been discussed
extensively, see refs [8, 9] for example.) The solitons are envelope solitons, that is, the envelope
function of a wavepacket has a soliton shape.
The experimental arrangement is shown schematically in Fig. 1. We studied spin-wave wave-
packets travelling in a single-crystalline (111)-oriented yttrium iron garnet (YIG) film of 7 µm
thickness, 1.5 mm width, and 40 mm length, forming a waveguide. Two short-circuited microstrip
antennae (length 2.5 mm, width 0.05 mm) for the excitation and detection of spin waves are at-
tached to the film at a distance l = 8 mm apart from each other. To compensate for the damping
of the waves and to maintain uni-directionality of the ring, a linear electronic amplifier connects
the output and input antenna. Its gain is chosen such that the system is below the self-generation
threshold for chaotic behavior when the phase-sensitive amplifier discussed below is disabled. An
external magnetic field of 1870 Oe is applied along the propagation direction, thus allowing for the
propagation of packets of “backward volume” spin waves.
In the linear regime, that is, for low excitation amplitudes, the ring shows a series of resonant eigen-
modes with frequencies in the interval 7.25 − 7.30 GHz, spaced by ∆ν = 3.2 MHz, and wavevectors
k < 100 − 150 cm−1 . The exact frequencies of the modes are determined by the phase-quantization
condition over the entire ring: k · l=2πn, with integer n. An estimation shows that the phase shift
accumulated in the electronic components of the ring is negligibly small. The time for a full circuit
around the ring is T0 = 310 ns.
It is necessary to have a selection mechanism to individually address each eigenmode of a nonlinear
ring. One widely employed method is the use of a critical threshold mechanism: only the mode
with the largest amplification gain is generated, at the expense of all others. This is commonly
employed in, for example, a ring laser system. Other reported methods are the use of frequency
filters or time gate filters (active mode locking) [10–13]. As a very versatile approach for mode
selection, we propose and demonstrate the use of a phase-selective amplifier. This amplifier only
allows modes to exist on the active ring, which have a well-defined preset phase at the position of
the amplifier in the ring. In conjunction with a simple time-gating filter and phase-sensitive mode
detection, this approach provides an elegant way to study the nonlinear eigenmodes in the active
ring.
To achieve phase-selective amplification we use the technique of parametric parallel pumping.
A sufficiently strong microwave field with frequency 2ν can parametrically amplify a spin-wave
packet of frequency ν [14, 15]. The pumping microwave field is sent to a microstrip resonator
attached to the film centered between the two antennae as shown in Fig. 1. The width of the
resonator, which determines the area of pumping, is d = 0.05 mm. Since k · d 2π, non-adiabatic
AG Magnetismus
39
TU Kaiserslautern
pumping is achieved and results in a phase-sensitive amplification coefficient: waves having phases
π/4 + nπ with respect to the pumping field are amplified most efficiently (see ref. [16] for details).
To provide a useful definition of the phase of a soliton-like wavepacket, we refer to the case of a
one-dimensional spin-wave wavepacket propagating in an unconfined magnetic medium. The dy-
namic magnetization m(x,t) describing the packet can be written as m(x,t) = f (x−vgt) exp(iϕ(x,t))
with ϕ(x,t) = 2πνt − kx − ϕa (x), where f is the envelope function, vg is the group velocity, ν and
k are the carrier frequency and wavevector, respectively, and ϕa (x) is an additional phase chirp
due to nonlinear effects. It is known [1, 17] that for a soliton, ϕa (x) is almost constant across the
soliton peak. Therefore, one can characterize a propagating soliton by its phase as a whole, for
example, at the maximum point of the envelope function: ϕ(t) = ϕ(x,t) for x = vgt. This phase
can be detected by mixing the soliton pulse with a continuous-wave (c.w.) reference signal, as is
shown in Fig. 1. Note that as the group velocity of the waves under consideration differs consider-
ably from the phase velocity v ph = 2πν/k the defined phase ϕ is time-dependent, that is, it changes
constantly while the soliton is propagating.
Without pumping in the parametric amplifier, no nonlinear modes are observed. Pumping is ap-
plied as a sequence of microwave pulses with a duration t = 15 − 35 ns and period Tp . It was found
that stable excitations can be observed if Tp is commensurate with T0 : a deviation of only a few
nanoseconds destroys the mode. For brevity, the following representative results are presented and
discussed for Tp = T0 , T0 /2, and T0 /3.
We begin with the simplest case of one wavepacket in the loop, that is Tp = T0 . Figure 2(left pan-
els) illustrates the waveforms of the detected eigenmodes, with curve (a) showing the waveform of
the pumping field as a reference. By sweeping the frequency (or, equivalently, the applied mag-
netic field) in a narrow interval of 20 MHz (8 Oe), several nonlinear eigenmodes are subsequently
observed. Without phase analysis all modes detected at different frequencies are indistinguishable
and consist of a series of soliton-like pulses spaced in time by T0 (curve (b)). The phase detection
technique reveals two possible types of one-soliton modes.
First, we find solitons all having the same phase, as displayed in curve (c). We label them “N” for
“normal” solitons. Second, we find a second set of modes with the phase difference between two
consecutive solitons of π, as displayed in curve (d). These modes need to travel twice about the
ring to meet the initial condition. Because of their similarity to the Möbius strip, we label them
“M” for “Möbius” mode. The detected waveform of the M-mode can be reversed by changing the
c.w. reference phase by π (curve (e)). Both modes are observed alternatively at different carrier
frequencies νN and νM with νN − νM = 1.6 MHz, that is, ∆ν/2.
The physical nature of the N-modes is easily understood: a soliton is propagating about the ring
with the closed-loop phase shift of 2πn. This set of modes, each mode characterized by n, is the
nonlinear counterpart of linear excitations. The M-mode, however, has no analogue in the linear
case: after completing a single loop, the phase accumulation of the soliton is 2π(n + 1/2), and
the soliton needs to travel a second loop to achieve the initial state. We point out that the possible
existence of Möbius modes, that is eigenmodes with a phase shift per loop that differs from an
integer multiple of 2π, was not considered by Carr et al. [4, 5].
We also find eigenmodes consisting of several soliton-like wave-packets per loop. This happens
when the pumping period is smaller than T0 . Figure 2(middle panels) displays the case of Tp
= T0 /2, that is, a system with two equally spaced wavepackets per loop. In a manner similar
to Fig. 2(left panels), the top curve (curve (a)) demonstrates the pumping waveform, while the
other curves are the measured waveforms of the different modes detected by phase analysis. The
AG Magnetismus
40
TU Kaiserslautern
a b c
a)
a)
(000)N
b)
b) (00)N
7.2655
(000)M
N (0π)N
c) c)
7.2657
d)
M d) (00)M (0π0)N
7.2672
e)
(0π0)M
e) (00)N
7.2687
0 300 600 900 0 300 600 900 0 300 600 900

Time (ns) Time (ns) Time (ns)
Fig. 2: Waveforms of the modes observed in the ring. Left panels: Tp = T0 , one-soliton modes. Curve
(a): waveform of the pumping field; curve (b): waveform of a nonlinear mode obtained without phase-sensitive
detection; curves (c) and (d): waveforms of two different modes detected using phase-sensitive detection; curve
(e): the same as curve (d) with the phase of the c.w. shifted by π. Middle panels: Tp = T0 /2, two-soliton modes.
Curve (a): waveform of the pumping field; curves (b) - (e): waveforms of different modes with the numbers
being the frequency of each mode in units of GHz. Right panels: Tp = T0 /3, three-soliton modes; waveforms
of the pumping field and of the observed modes. The mode nomenclature is explained in the text.
numbers adjacent to each curve indicate the corresponding values of the mode frequencies in units
of GHz. The three vertical dashed lines indicate the time interval T0 that each of the soliton packets
needs to complete the loop.
As for the one-soliton case discussed above, these modes can be divided into normal and Möbius
modes. We observe two normal modes (curves (b) and (c)) and one Möbius mode (curve (d)).
Although the two normal modes are not completely degenerate, their frequency separation is much
smaller than ∆ν. As found for the one-soliton case, the difference between the Möbius mode and
both the normal modes is νN −νM ∼ = 1.6 MHz = ∆ν/2. By sweeping the frequency, alternating nor-
mal and Möbius modes appear, as demonstrated by curve (e), which (with respect to the observed
signal) is identical to curve (b), but separated in frequency by the amount ∆ν.
As a further illustration, in Fig. 2(right panels), we demonstrate the case of three equally spaced
soliton-like pulses in the ring with a temporal separation, that is, pumping period of T0 /3. Two
normal modes and two Möbius modes are found. Again, normal and Möbius modes are observed
to alternate with the same frequency interval.
Thus the phase sensitivity of the parametric pumping process constitutes a phase selection rule for
the allowed eigenmodes of the ring. Owing to localized parallel pumping with frequency 2ν, the
pumping gain is at maximum for phases πn, allowing for modes with a phase shift per loop of 2πn
and 2π(n + 1/2). The former values correspond to normal modes and the latter to Möbius modes.
To classify the multi-soliton modes found, we quote the soliton phase relative to the phase of an
arbitrary chosen soliton, which we set to zero without loss of generality. A soliton having the same
AG Magnetismus
41
TU Kaiserslautern
phase will be indicated by “0”, and one with a phase difference of π by “π”. For Tp = T0 only two
one-soliton modes exist, a normal mode, (0)N, and a Möbius mode, (0)M. For the case of two-
soliton modes we find, using combinatorics, four possible combinations: (00)N, (0π)N, (00)M,
and (0π)M. However, the last two modes are physically identical, as they can be transferred into
each other by shifting the origin of the timescale by T0 /2.
Some of the N-eigenmodes are expected to have broken symmetry, since they cannot evolve
monotonously from their linear counterparts. We find that the (00)N mode has a linear counterpart,
while the (0π)N mode has the predicted broken symmetry. This is experimental verification of a
symmetry-breaking eigenmode, as predicted [4, 5]. A similar analysis for Tp = T0 /3 predicts the
existence of four non-equivalent three-soliton modes: (000)N, (0π0)N, (000)M, and (0π0)M. All
these modes are observed in the experiment, as demonstrated in Fig. 2.
To good approximation, the multi-soliton eigenmodes can be viewed as a set of soliton-like wave-
packets with weak interaction in the case of a small overlap. We conclude by briefly reporting an
experiment that unambiguously demonstrates this interaction. Using appropriate time gating we
prepared a mode consisting of two solitons with a variable time separation, which travel about the
ring. For simplicity, the pumping frequency is chosen in such a way that both solitons correspond
to normal modes, that is their closed-loop phase shift is 2πn. The relative phase of the first soliton
is set to zero. Owing to phase selection imposed by the parametric pumping process, the phase
of the second soliton can either be 0 or π. By reducing the time delay ∆T between the solitons
one can investigate their interaction. It was experimentally found that independently of the initial
phase difference between the solitons at large ∆T the phase difference between them for small
delays is always π. This fact provides clear evidence of a phase-dependent interaction between the
two solitons. The origin of the observed interaction is not clear so far and represents a challenge
for future studies.
A full report will be published in Nature [18].
Support by the Deutsche Forschungsgemeinschaft is gratefully acknowledged.
References
[1] M. Remoissenet, Waves Called Solitons: Concepts and Experiments (Springer-Verlag, Berlin, 1966).
[2] L. Khaykovich et al., Science 296, 1290 (2002).
[3] G. P. Agrawal, Nonlinear Fiber Optics (Academic, San Diego, 1995).
[4] L.D. Carr, C.W. Clark, W.P. Reinhardt, Phys. Rev. A 62, 063610 (2000).
[5] L.D. Carr, C.W. Clark, W.P. Reinhardt, Phys. Rev. A 62, 063611 (2000).
[6] X. Liu, K.J. Qian, F.W. Wise, Phys. Rev. Lett. 82, 4631 (1999).
[7] M. Bauer et al., Phys. Rev. Lett. 81, 3769 (1998).
[8] Yu.S. Kivshar, B.A. Malomed, Rev. Mod. Phys. 61, 763 (1989).
[9] Z. Li, L. Li, H. Tian, G. Zhou, K.H. Spatschek, Phys. Rev. Lett. 89, 263901 (2002).
[10] B.A. Kalinikos, M.M. Scott, C.E. Patton, Phys. Rev. Lett. 84, 4697 (2000).
[11] B.A. Kalinikos, N.G. Kovshikov, M.P. Kostylev, H. Benner, JETP Lett. 76, 310 (2002).
[12] B.A. Kalinikos, N.G. Kovshikov, C.E. Patton, Phys. Rev. Lett. 80, 4301 (1998).
[13] G. Steinmeyer, D.H. Sutter, L. Gallmann, N. Matuschek, U. Keller, Science 286, 1507 (1999).
[14] A.V. Bagada, G.A. Melkov, A.A. Serga, A.N. Slavin, Phys. Rev. Lett. 79, 2137 (1997).
[15] P.A. Kolodin et al., Phys. Rev. Lett. 80, 1976 (1998).
[16] G.A. Melkov et al., Phys. Rev. E 63, 066607 (2001).
[17] J.M. Nash, P. Kabos, R. Staudinger, C.E. Patton, J. Appl. Phys. 83, 2689 (1998).
[18] S.O. Demokritov, A.A. Serga, V.E. Demidov, B. Hillebrands, M.P. Kostylev and B.A. Kalinikos, Nature 426,
159 (2003).
AG Magnetismus
42
TU Kaiserslautern
6.5 Black soliton formation from phase-adjusted spin wave packets
A.A. Serga, A. Andre, S.O. Demokritov, and B. Hillebrands1
An envelope dark soliton can be understood as a stable propagating dip in a carrier wave of other-
wise constant amplitude in a nonlinear medium. The characteristic feature of such a dark soliton
is a nonzero jump in the phase of its carrier wave as one crosses the dip [1, 2]. In the case of a
black soliton, when the maximum depth of its dip is equal to the amplitude of the carrier wave, this
phase shift is exactly equal to 180◦ .
The number of the formed dark solitons, as well as their amplitude and phase characteristics,
strongly depend on the initial conditions, i.e., on the amplitude and phase modulation of the carrier
wave before it enters the nonlinear medium. In a magnetic system, as studied here, so-called
microwave magnetic envelope (MME) dark solitons evolve from an initial gap created in the carrier
spin wave without any special phase modulation [3,4]. In this case only multiple solitons should be
formed, so the phase jumps in successive solitons are compensated, and the net phase shift across
the entire output signal is equal to an integer multiple of 360◦ . However the process of soliton
formation can be strongly modified by the diffrent nonlinear phenomena.
Here we report on a new approach to create single dark and black MME solitons as well as on
the first experimental observation of a so-called excitation-induced nonlinear phase shift. This
phase shift is created by the gradual entrance of a relatively slow spin wave into the nonlinear
magnetic medium and is proportional to the duration of the input dark pulse and to the power of
its carrier wave (see Eqs. (2), (3) in Ref. [5]). A spin wave pulse of duration t0 will accumulate a
phase shift since its leading edge propagates for the time t0 longer in the nonlinear medium than
its trailing edge. We show that the pre-compensation of this phase shift is a necessary condition
for the formation of a single black MME soliton with a 180◦ phase jump.
Figure 1 shows a schematic diagram of the experimental setup. Its main part is a spin-wave wave-
guide, which consists of a strip of a single-crystal yttrium-iron-garnet (YIG) film on a gadolinium
gallium garnet substrate and a pair of short-circuited microstrip antennae. The YIG film shown by
Oscilloscope
out
Phase G
Shifter Low Noise
Amplifier H0
MW MW
Source Switch
in
Phase MW
Power YIG film
Shifter Switch
Amplifier delay line
Fig. 1: Schematic diagram of the microwave (MW) experimental setup.
1 In collaboration with A.N. Slavin, Department of Physics, Oakland University, Rochester, Michigan, USA.
AG Magnetismus
43
TU Kaiserslautern
Fig. 2: Normalized output sig-

nals formed from the input pulses
with 6 ns gap. Frames (a) and (b)
correspond to the input power of
0.2 mW (linear case) and 28 mW
(nonlinear case). In (b) the
dashed line shows the profile of
a dark soliton obtained by us-
ing the parameters as in (a) apart
from the larger input power. The
solid-line profile corresponds to
the black soliton obtained after
phase adjustment by 50◦ , as de-
scribed in the text. The inset in
(b) shows the interference signal
resulting from interference of the
output black soliton signal with
a monochromatic reference sig-
nal (solid line) together with the
black soliton (dotted line). The
180◦ phase jump coincides with
the position of the soliton dip.
a shaded rectangular in Fig. 2 has the following parameters: length, width, thickness are 18 mm,
1 mm, and 5 µm, respectively; saturation magnetization: 4πMs = 1750 G. The 10 GHz ferromag-
netic resonance linewidth 2∆H is 0.55 Oe. The film has unpinned surface spins.
The antennae are situated transversely to the long side of the YIG sample and are separated by
8 mm. One of the antennae is used for the excitation of spin waves and the other for their detection
after travelling the distance between the antennae. The antennae’s width and length are 50 µm and
2.5 mm, respectively.
The sample was magnetized by a static magnetic field of H0 = 1745 Oe applied in the film plane
and transversely to the long edge of the strip, see Fig. 1. The input SW signal was exited by
an electromagnetic wave with a carrier frequency of ωs = 2π · 6.950 GHz. In this case, the input
antenna excites magnetostatic surface waves (MSSW) with carrier wave number ks = 85 rad/cm.
For unpinned films these MSSW have a monotonic dispersion curve ω(k) with both negative dis-
persion, D = ∂2 ω/∂k2 , and negative nonlinearity, N = ∂2 ω/∂|u|2 , where u is the amplitude of the
wave. As the positive product ND > 0 allows for dark type solitons [1–3, 5], the MSSW are able
to form dark envelope solitons while travelling between the antennae.
In order to vary the initial phase shift over the gap in the carrier wave, the input signal was com-
posed of two coherent electromagnetic microwave pulses with controlled phase shift between them
in the range (0, 2π) (see Fig. 1 and Fig. 2a). The initial gap duration T was varied in the range from
5 ns to 15 ns by changing the delay time between the pulses. The aggregate signal was amplified by
a power amplifier with controlled gain and sent to the input antenna of the spin wave delay line (1).
The input peak power Pin was varied in the range from 0.2 mW up to 50 mW. The total duration
of the aggregate input signal was restricted to 100 ns to avoid additional nonlinear processes. For
example, a further increase of Pin and/or the pulse duration leads to a pronounced suppression and
AG Magnetismus
44
TU Kaiserslautern
distortion of the pulses due to four-magnon decay into exchange dominated short-wavelength spin
waves.
To analyze the phase jump inside the output signal we measured the signal obtained by interference
of the output signal with a phase adjusted reference signal from the microwave source (Fig. 1).
First, we discuss the linear case, i.e., Pin = 0.2 mW. The shape of the output signal is shown in
Fig. 2a. Due to the distortions created by the limited frequency band of the delay line and by the
spin wave dispersion in the film, the two pulses forming the output signal have bell-like profiles and
the initially rectangular T = 6 ns wide gap spreads up to 28 ns (measured at half-depth level). As
a result of a slight dispersion-induced phase modulation of the carrier wave the lowest amplitude
in the dip was found to be non-zero. A proper adjustment of the phase shift in the input pulse
sequence brings the two delayed pulses back to the case of destructive interference in the region
of their overlap. Thus, the amplitude in the dip becomes almost exactly equal to zero. The actual
value of this adjustment of the initial phase depends on the gap width T .
An increase of the input power Pin results in the crossover from the linear to the nonlinear regime
of spin wave propagation. The nonlinear regime causes two simultaneous changes in the shape
of the output signal: the central dip of it becomes narrower and the minimum amplitude in the
dip increases as shown by the dashed curve in Fig. 2b. This narrowing provides a clear evidence
of the dark soliton formation [3]. The threshold power of this process is about 7 mW. We note,
that the two maxima surrounding the dip in the output pulse demonstrate a slight broadening with
increasing Pin , because the nonlinear distortions add to the dispersion spreading of pulsed signals
in the case of ND > 0
The detected reduction in the depth of the dark soliton dip is a consequence of the nonlinearity-
induced spatial phase modulation of the carrier MSSW. If some phase shift of opposite sign is
introduced into the two pulses comprising the input signal, the depth of the dip tends to become
zero, while the phase jump over the dip in the output pulse tends to equal the initial 180◦ value.
Thus, the measurement of the value of the compensating phase shift introduced by the external
phase shifter (Fig. 1) provides us with information about the intrinsic excitation-induced nonlinear
phase modulation.
Figure 3 shows the dependence of the measured induced phase shift on the input power Pin for two
gap widths T1 = 6 ns and T2 = 15 ns, respectively. In accordance with theoretical predictions [5] the
induced phase shift φ(u) ∝ C/T + |u|2 NT (see Eq. (2) in Ref. [5]) increases with increasing power
(∝ |u|2 ) of the excited spin waves, while the dependence on T is more complicated. This behavior
is clearly seen for the values of the input power above the threshold of dark soliton formation
(Pin > 7 mW). For the smaller values of the input power (i.e. in the linear region) the accuracy of
our phase measurements is rather low, which prevents us from making a quantitative comparison
of our experimental data with the results of the theory [5] in this linear region.
If the input power exceeds 35 mW for T = 15 ns the phase shift starts to saturate. This is ac-
companied by the appearance of distortions in the shape of the output pulse caused by the above
mentioned process of parametric excitation of short exchange-dominated spin waves.
The above described phase adjustment technique enables us to observe the formation of a single
black spin wave envelope soliton under well-determined circumstances. For example, adjusting the
phase shift in the input pulse sequence we were able to maximize the depth of the initially created
dark soliton in such a way that the amplitude at the dip minimum is almost exactly zero (see Fig. 2),
and a pronounced single black soliton state was achieved for Pin = 28 mW and T1 = 6 ns. The width
of the obtained black soliton is comparable with the width of the originally formed dark soliton
AG Magnetismus
45
TU Kaiserslautern
90 Fig. 3: Excitation-induced non-

80
Width of the gap in the input signal: linear phase shift across the dip
6 ns in the carrier spin wave as a func-
Induced phase shift (deg)
70 15 ns tion of the input power.

60
50
40
30
20
10
0
0 10 20 30 40
Input power (mW)
having non-zero amplitude at the minimum of its dip.

The inset in Fig. 2b shows the interference of the output pulse, corresponding to this black soliton,
and the monochromatic reference signal. The transition from destructive to constructive interfer-
ence marks the pronounced 180◦ phase jump inside of the black soliton. When Pin is increased
above 28 mW the soliton appearance becomes even more pronounced: the dip becomes narrower
and steeper. However, in this strongly nonlinear regime it is difficult to separate the influences
of the different nonlinear processes (formation of a soliton and excitation of exchange-dominated
spin waves), and the picture becomes less clear.
To summarize, the formation of a single dark spin wave envelope soliton was experimentally real-
ized for long-wavelength surface magnetostatic spin wave using as an input signal a sequence of
two microwave radio-pulses with controlled gap and phase shift between them. The variation of
the depth of the generated dark soliton and the creation of the fundamental black soliton state were
achieved via manipulating the phase shift between the pulses in the input sequence. The excitation-
induced nonlinear phase shift in the output dark pulse (soliton) was experimentally observed. The
artificial pre-compensation of the nonlinear phase shift enables us to observe the generation of a
single fundamental black spin wave envelope soliton having a 180◦ phase jump.
This work was partially supported by the Deutsche Forschungsgemeinschaft, by the National Sci-
ence Foundation, Grant No. DMR-0072017, and by the Oakland University Foundation.
References
[1] M. Remoissenet, Waves Called Solitons: Concepts and Experiments (Springer-Verlag, Berlin, 1966).
[2] A.N. Slavin, IEEE Trans. Magn. 31, 3479 (1995).
[3] M. Chen, M.A. Tsankov, J.M. Nash, C.E. Patton, Phys. Rev. Lett. 70, 1707 (1993).
[4] J.M. Nash, P. Kabos, C.E. Patton, R. Staudinger, J. Appl. Phys. 79, 5367 (1996).
[5] A.N. Slavin, Yu.S. Kivshar, E.A. Ostrovskaya, H. Benner, Phys. Rev. Lett. 82, 2583 (1999).
AG Magnetismus
46
TU Kaiserslautern
6.6 Parametric generation of forward and phase-conjugated spin-wave bul-

lets in a magnetic film
A.A. Serga, S.O. Demokritov, and B. Hillebrands1
Although predicted for light [1], wave bullets − stable two-dimensional nonlinear wave packets
self-focused in space and time − were first experimentally observed in the system of spin waves
in yttrium-iron garnet (YIG) films [2]. The experimental investigation of wave bullets in magnetic
films has the advantage of direct two-dimensional access to the medium and of the possibility to
follow the bullet formation and propagation in real time due to its relatively low speed (compared
to the speed of light) using the space- and time-resolved Brillouin light scattering technique [3].
An additional advantage is the fact that in garnet films it is possible to realize an effective para-
metric interaction of a propagating spin-wave packet with non-stationary (pulsed) electromagnetic
pumping [4].
Here, in contrast with the traditional method of spin-wave bullet formation from an initially nonlin-
ear input spin wave packet directly excited using a microwave antenna, we report the experimental
generation of two-dimensional spin wave bullets in both amplified and phase conjugated reversed
wave packets formed as a result of parametric interaction of an initial spin wave packet with elec-
tromagnetic pumping of double frequency.
The experiment utilized a spin-wave waveguide situated on top of a microstrip structure as shown
in Fig. 1, which was designed for transformation of an electromagnetic signal into spin waves and
vice versa. Two microstrip antennae of 25 µm width and 2 mm length spaced 8 mm apart were
lifted 100 µm up above the surface of the structure to avoid the direct contact of the YIG film with
the other parts of the circuit and, as a result, to eliminate any undesirable excitation of parasitic
spin waves by the microstrip feeders. One of the antennae was used for excitation of spin waves,
while the second was used for real time monitoring of the propagated spin wave signal.
The wide (4.1 × 30 mm2 ) spin-wave waveguide was cut out from a single crystal YIG(111) film
with thickness of 5 µm, grown on a gallium-gadolinium garnet substrate. The film’s saturation mag-
Pout
Microstrip antenna Fig. 1: Experimental setup.
Pp
AMPLIFIED
Probe laser beam SPIN
WAVE
Microstrip
Pin YIG film
antenna REVERSED
SPIN WAVE waveguide
INITIAL
Z
H0
SPIN WAVE
Y
Dielectric
Inelastically scattered light resonator
1 In collaboration with A.N. Slavin and P. Wierzbicki, Department of Physics, Oakland University, Rochester,
Michigan, USA, as well as with V. Vasyuchka, O. Dzyapko, and A. Chumak, Department of Radiophysics,
Taras Shevchenko National University of Kiev, Kiev, Ukraine.
AG Magnetismus
47
TU Kaiserslautern
netization is 4πMs = 1750 G and the full linewidth of ferromagnetic resonance is 2∆H = 0.6 Oe.
The sizes of the waveguide were chosen to be larger than the sizes of the antennae, the distance
between them, and the characteristic length of non-linear packet formation [5, 6].
A rectangular dielectric resonator of a special shape was developed to create a pumping mag-
netic field inside the YIG film. The resonator sizes are 1 × 9.45 × 5.85 mm3 . The waveguide was
mounted in a thin (0.8 mm) slot in the center of the resonator, where the intensity of the rf magnetic
field of the used resonator mode TE11δ has its maximum. In this case the pumping magnetic field
is applied along the long side of the waveguide and parallel to the direction of spin wave propaga-
tion. The resonance frequency and loaded quality factor were equal to f p = 7.932 GHz and 184,
respectively.
The bias magnetic field H0 was applied along the z axis (see Fig. 1), i.e., in the plane of the film
and perpendicular to the antennae. Thus, as the thin microstrip antenna excites efficiently only
dipolar dominated spin waves propagating perpendicularly to the antenna’s axis, the geometry of
the experiment corresponds to magnetostatic backward volume waves (MSBVW). These waves
are able to form envelope bullets. Note here that in this geometry the rf pumping magnetic field
coincides with the direction of the bias field, and thus the case of a parallel pumping is realized [7].
The MSBVW packets were excited by the rectangular electromagnetic pulses of duration 30 ns
with carrier frequency fs = 3.966 GHz and varying power Pin . For the applied bias field H0 =
800 Oe the excited MSBVW packets have a carrier wave number of kz = 95 rad/cm.
1.4
Incident packet
Dielectric resonator area
Amplified packet
1.2
Propagation Propagation
Width (mm)
1.0 direction direction

1
0.8 1
2
0.6 2
3
0.4 3
0.2 (a) (b)
0.0 0.5 1.0 1.5 2.0 2.5 3.51.5
1.0 4.02.0 2.5 3.0 3.5 4.0 4.5
Distance (mm) Distance (mm)
1.4 1,2
Reversed packet
Dielectric resonator area
1
Width and length (mm)
Intensity
1.2 1,0
Propagation
Intensity (a.u.)
Width
Width (mm)
1.0 direction 1
0,8
0.8 2
3 0,6
0.6
0,4
0.4
Length 0
(c) 0,2 (d)
0.2
0.0 0.5 1.0 1.5 2.0 2.5 3.5
0 4.0
50 100 150 200 250
Distance (mm) Time (ns)
Fig. 2: Changes of spin-wave packet characteristics with propagation distance z and time t. Evolutions of the
transversal width Ly of the incident, amplified and reversed spin wave packets are shown in Frames (a), (b), and
(c) respectively. Curves (1), (2), and (3) correspond to the input powers Pin of 22 mW, 146 mW, and 226 mW,
respectively. Frame (d) displays the time evolutions of the width Ly and length Lz , as well as of the pulse’s
amplitude for the amplified spin wave pulse with Pin = 226 mW.
AG Magnetismus
48
TU Kaiserslautern
Using the space- and time-resolved BLS technique [3] two-dimensional maps of the spin wave
intensity I(y, z), each sampled over a time interval of 1.7 ns, were recorded. Using these maps
the mean packet position (z0 (t), y0 (t)) and the transverse packet width Ly as well the longitudinal
length Lz can be easily determined as functions of the propagation time and distance [8].
The pumping pulse with duration of 39 ns and peak power 30 W was applied when the spin-wave
packet passes through the dielectric resonator. As a result the maximum amplification of the de-
layed pulse observed on the output antenna reached 27 dB. Simultaneously with the amplification
of the input spin-wave packet the reversed spin wave packet propagated in backward direction
with respect to the initial packet was formed by strongly amplified thermal spin waves. The am-
plitude of these packets depends both on the power Pin of the input pulse and the pumping power.
To avoid overheating of the sample around the resonator as well as other uncontrolled effects the
pumping power and duration were fixed and the amplitudes of the amplified and reversed pulses
were controlled only by the variations of Pin in the range from 2.2 mW to 700 mW.
First, let us consider spin-wave bullet formation from a spin wave packet on the way to the res-
onator. The evolution of the incident packet as a function of the input power 2.2 ≤ Pin ≤ 22 mW is
characterized by a monotonic widening of the packet (see curve 1 in Fig. 2a). A further increase
of the input power up to 226 mW leads to a pronounced nonlinear compensation of the transver-
sal widening of the packet followed by spin-wave bullet formation. Increasing Pin above 500 mW
results in strong transversal compression of the spin wave packet followed by its fast collapse and
destruction. A systematic oscillatory behavior of Ly shown in Fig. 2a is caused by the interference
of a few spin-wave width-modes in the YIG waveguide [9].
As a result of the parametric interaction with the microwave magnetic field in the region of the
dielectric resonator the propagating spin wave packet increases its length and width. Just behind
the resonator (as it is seen in Fig. 2b) the widths Ly both of the previously linear (curve 1) and of the
nonlinear (curves 2-3) spin wave packets are practically identical. However, the further behavior
of the amplified spin wave packet depends on the power. At the same time the main features of
the packets’ evolution are common for all input powers Pin , as seen in Fig. 2b. First, the amplified
packet is essentially transversally compressed. Second, the compressed packet keeps its width for
a long propagation path. For example, in the case of Pin = 226 mW the propagation distance, where
the packet width Ly is practically constant, is about 1.5 mm (marked with two vertical arrows in
Fig. 2b).
As it is seen in Fig. 2d the described process of the transversal narrowing is accompanied by
a simultaneous decrease of the spin-wave packet length Lx and increase of the packet intensity,
which is due to the strong spatial concentration of the spin wave energy.
Thus, the process of spin-wave packet compression has a two-dimensional character and leads to
the formation of the stable two-dimensional nonlinear spin wave excitation, the spin wave bullet.
As the input power Pin increases the compression process develops faster and the compression
degree rises (Fig. 2b). An increase of Pin above 300 mW results in a wave collapse and breaking of
the packet into parts.
The behavior of the reversed spin wave packet is even more intriguing. Due to two-dimensional
phase-conjugation by means of parametric interaction, a distinct transmission narrowing is ob-
served even for quasi-linear packets (see curve 1 in Fig. 2c). Comparing the slopes of the curves 1
in Fig. 2a and Fig. 2c one can conclude, that the above effect is the counterpart of the diffraction
broadening of the initial spin wave packet. An increase of Pin involves an additional nonlinear
self-focusing of the spin wave packet in the process. It is remarkable that not only the compression
AG Magnetismus
49
TU Kaiserslautern
develops faster, but, as it is shown in Fig. 2, the strongly compressed reversed spin wave packet
moves as a solitary stable two-dimensional excitation. The broadening of this nonlinear reversed
spin wave packet (see curve 2 in Fig. 2c) is much slighter in comparison with its previous compres-
sion and is caused by the diffraction which plays a more and more important role with dissipation
of the spin-wave packet energy and the diminishing influence of the nonlinearity. Thus, Fig. 2c
clearly demonstrates that there are profound regions of propagation distances (or times) where
the transverse sizes of the reversed wave packets are almost constant, i.e., where propagation of
self-generated bullets is observed.
Thus, for the first time we have observed bullet formation both by parametrically amplified spin
wave packets and by the phase conjugated spin waves parametrically generated from the thermal
level.
As the characteristic behavior of the observed bullets does not relay on the method of the initial
excitation of the waves packets they can be classified as a two-dimensional natural excitation of a
nonlinear medium with dispersion, diffraction and dissipation.
This work was supported by the Deutsche Forschungsgemeinschaft, by the U.S.A. National Sci-
ence Foundation (Grants No. DMR-0072017 and INT-0128823).
References
[1] Y. Silberberg, Opt. Lett. 15, 1282 (1990)
[2] M. Bauer, O. Büttner, S.O. Demokritov, B. Hillebrands, V. Grimalsky, Yu. Rapoport, A.N. Slavin, Phys. Rev.
Lett. 81, 3769 (1998).
[3] S.O. Demokritov, B. Hillebrands, A.N. Slavin, Phys. Reports 348, 441 (2001).
[4] G.A. Melkov, A.A. Serga, A.N. Slavin, V.S. Tiberkevich, A.N. Olejnik, A.V. Bagada, JETP 89, 1189 (1999).
[5] O. Büttner, M. Bauer, S.O. Demokritov, B. Hillebrands, Yuri S. Kivshar, V. Grimalsky, Yu. Rapoport,
A.N. Slavin, Phys. Rev. B 61, 11576 (2000).
[6] R.A. Staudiner, P. Kabos, H. Xia, B.T. Faber, C.E. Patton, IEEE Transactions on Magnetics 34, 2334 (1998).
[7] A.G. Gurevich G.A. Melkov, Magnetic Oscillations and Waves (CRC, New York, 1996).
[8] A.A. Serga, S.O. Demokritov, B. Hillebrands, A.N. Slavin, J. Appl. Phys. 93, 8758 (2003).
[9] O. Büttner, M. Bauer, C. Mathieu, S.O. Demokritov, B. Hillebrands, P.A. Kolodin, M.P. Kostylev, S. Sure,
H. Dötsch, V. Grimalsky, Yu. Rapoport and A.N. Slavin, IEEE Trans. Magn. 34, 1381 (1998).
AG Magnetismus
50
TU Kaiserslautern
C. Magnetic Films and Double Layers

6.7 Surface smoothing and reduction of Néel “orange peel” coupling for
magnetic tunnel junctions using low energy argon ions
P.A. Beck, B.F.P. Roos. S.O. Demokritov, and B. Hillebands
There is great interest in improved preparation methods for ultrathin metallic films driven by
applications such as giant magneto-resistance (GMR) elements and magnetic tunnel junctions
(MTJ) [1]. MTJs consist typically of a sandwich structure with a pinned ferromagnetic layer
(typically Co or CoFe) and a free ferromagnetic layer (typically permalloy, NiFe) separated by an
insulating layer. For usage in applications such as magnetic random access memory (MRAM) or
read heads in hard disk drives, a small and reproducible switching field for the free layer is strongly
desired. The morphology of the interface has a great influence on the switching field. It was Néel
who first considered magnetic dipole coupling between the two layers due to rough interfaces, so-
called “orange peel coupling” [2]. Schrag et al. have shown that magnetostatic and Néel coupling
are responsible for the coupling between the pinned and free magnetic layer in MTJs [3, 4].
Not only the amplitude of the roughness but also its lateral scale determine the strength of the
Néel coupling. If a coherent sinusoidal roughness profile with amplitude h and wavelength λ is
assumed, the coupling field HN acting on the free layer in a “pinned layer/barrier/free layer” stack
is given by
√
π2 h2 −2π 2 tb
HN = √ Ms exp , (1)
2 λtf λ
where tf is the thickness of the free layer, tb that of the barrier, and Ms the magnetization of the
pinned layer. At given√h the coupling field increases with increasing wavelength λ, reaches its
maximum at λmax = 2π 2tb , and then it exponentially decreases. Thus the Fourier components of
the interface roughness near λmax , which is about 10 nm for a typical MTJ, are of largest importance
for the Néel coupling mechanism.
The as-grown interface roughness is determined by the thermodynamics and kinetics of the growth
process and the deposition technique, and it cannot be reduced beyond certain limits. Here we
report the realization of a successful post-fabrication smoothing process based on low energy argon
ions. In agreement with theoretical models the coupling field measured on MTJs is largely reduced.
The stacks are prepared at room temperature in a multi-chamber molecular beam epitaxy system
with a base pressure of < 10−10 mbar using a five-pocket electron beam evaporator. The surface
morphology is studied by in-situ scanning tunneling microscopy (STM). For the magnetic studies
ex-situ magneto-optic Kerr effect (MOKE) magnetometry is used. All samples are prepared on
thermally oxidized Si(100) substrates with an 100 nm thick SiO2 layer. For the STM studies a
single 15 nm thick Py layer was used. The magnetic properties are determined for full MTJs
made by depositing 2.6 nm Al2 O3 /3 nm CoFe/2 nm Cr onto the Py layer. The Al2 O3 barrier layer
is prepared by plasma oxidation of a metallic Al film [5] using the same plasma source as for
surface smoothing. The surface smoothing process has been performed in normal incidence in a
separate reactor chamber with base pressure < 10−9 mbar, with sample transfer without breaking
AG Magnetismus
51
TU Kaiserslautern
Fig. 1: STM images of the surface morphol-

ogy of a 15 nm thick Ni19 Fe81 film on a SiO2
substrate: a) before ion bombardment and b-
d) after irradiation with argon ions with an
ion dose of 8 · 1016 ions/cm2 . The ion en-
ergy is b) 30 eV, c) 50 eV and d) 70 eV.
the vacuum. The chamber is equipped with an inductively coupled radio frequency (13.56 MHz)
plasma beam source providing a quasi-neutral Ar+ plasma beam. The ion energy can be chosen
in the range of 20 − 90 eV with a dispersion of less than ±5 eV. The current density is typically
50 − 70 µA/cm2 .
The surface roughness of a thin film can be roughly characterized by characteristic height and lat-
eral structure parameters. As roughness-height parameters the so-called peak-to-valley roughness
(rpv ) and the root-mean-square roughness (rrms ) are commonly used. They are directly obtained
from the analysis of the STM images. Additionally a lateral correlation length ξ can be deter-
mined from the height-height correlation function [6]. Figure 1a shows a typical STM image of
an as-grown 15 nm thick Py film. Characteristic are small grains with an average size in the or-
der of 10 nm. From the analysis of the images the following values are derived: rrms = 0.45 nm,
ξ = 45 nm. Figure 1b-d shows the Py surface after irradiation by Ar ions of different energies but
with the same ion flux of 8 · 1016 ions/cm2 . It illustrates the change in the surface morphology by
the irradiation process. The derived parameters are summarized in Fig. 2, which shows that both
rpv and rrms are drastically decreased by the irradiation process. For 70 eV the value of rrms drops
below 0.3 nm. On the contrary the irradiation causes an essential increase of ξ by a factor of more
than 100%. Thus, the presented experimental data indicate a significant surface smoothing effect
Fig. 2: Ion energy dependence of the

surface properties of a 15 nm thick
Ni19 Fe81 film on a SiO2 substrate:
a) Peak-to-Valley roughness , b) root
mean square roughness, c) correlation
length.
AG Magnetismus
52
TU Kaiserslautern
Fig. 3: Power spectra of the surface morphology of a 15 nm thick Ni81 Fe19 film on SiO2 obtained by a Fourier
analysis of the STM images.
by the irradiation with low energy argon ions.

For a more detailed characterization of the surface smoothing effect the Fourier power spectra of
the STM images are analyzed. Figure 3 shows Fourier power spectra of several STM images.
For each image the spectrum was calculated for a single line with a following average over all
parallel lines of the image. The shown Fourier spectra clearly demonstrate the smoothing effect
for different lateral scales. The smoothing is observed for all values of the analyzed lateral scales,
but is most pronounced for λ = 5 − 40 nm, which is in agreement with the data shown in Fig. 2c.
As it is discussed above, 5 − 50 nm is the interval of the lateral length components of the roughness,
which mainly contribute to the Néel interlayer coupling for typical barrier widths of 0.5 − 2 nm.
As seen in Fig. 3, even 30 eV ions significantly reduce the roughness with a lateral characteristic
length of 5 − 40 nm. It is more pronounced for 50 eV ions. The strongest effect is observed for
70 eV ions, which provide a smoothing in a broad range from 5 nm to 100 nm. One can conclude
that structures in the roughness with a characteristic lateral length scale below 10 nm are almost
completely suppressed.
Unfortunately, smoothing using ions with an energy exceeding a certain threshold, which typically
is 50 eV, may cause also undesirable effects, such as the creation of defects. To study this, the im-
pact of surface smoothing on magnetic interlayer coupling is investigated by studying the magneti-
zation loops of full MTJs. Figure 4 shows the magnetization loops for 15 nm NiFe/2.6 nm Al2 O3 /
3 nm CoFe/2 nm Cr MTJs, based on Py films with different surface roughness prepared by apply-
ing the smoothing process. The magnetization loop of a single 3 nm CoFe/2 nm Cr layer is also
presented in Fig. 4 for comparison. One can clearly see the switching of the two magnetic layers
in different fields. Due to a smaller coercive field of NiFe it is switched first in a weak field. After
this the ferromagnetic “orange peel” coupling between the NiFe and CoFe layers trends to shift the
switching field of the CoFe layer to weaker fields with respect to that of a single CoFe layer. Thus,
increasing the switching field of the CoFe layer in MTJs based on the smoothed NiFe layers can
be considered as the experimental confirmation of the reduced coupling. The reduced coupling,
as it is also seen in Fig. 4, causes an increase of the switching field of the NiFe layer. The effect
is not so strong as it is for the CoFe film due to the much larger thickness of the NiFe layer, but
it demonstrates the symmetry of the coupling. In agreement with the STM data, Fig. 4a and 4b
AG Magnetismus
53
TU Kaiserslautern
Fig. 4: Magnetization reversal loops measured with MOKE magnetometry for two different ion doses and for
a) 30 eV, b) 50 eV, and c) 70 eV ion energies for Si/SiO2 /15 nm Ni81 Fe19 /2 nm Al2 O3 /3 nm Co90 Fe10 /2 nm Cr.
For reference, loops of a as-grown sample and of a single free 3 nm Co90 Fe10 /2 nm Cr stack on SiO2 is shown
in all graphs.
show that irradiation with 50 eV ions causes a stronger reduction of the coupling, as it is detected
for 30 eV ions. However, a further increase of the ion energy up to 70 eV does not decrease the
coupling. Instead, the remagnetization loop of CoFe layer looses its squareness. The origin of this
effect is not clear up to now, although defect generation is likely.
In conclusion, we have found that irradiation with 50 eV argon ions of relatively low fluences of
1016 − 1017 ions/cm2 causes an essential smoothing effect for NiFe layers, the roughness on a
typical length scale of 5 − 40 nm being mostly reduced. The reduced roughness clearly manifests
itself in a reduced Néel “orange peel” coupling, measured on realistic MTJ structures, prepared on
the smoothed NeFe layers.
Financial support by the BMBF is gratefully acknowledged.
References
[1] J.S. Moodera, L.R. Kinder, T.M. Wong, R. Meservey, Phys. Rev. Lett. 74, 3273 (1995).
[2] L. Néel, Comptes Rendus Acad. Sci. 255, 1676 (1962).
[3] B.D. Schrag, A. Anguelouch, S. Ingvarsson, G. Xiao, Y. Lu, P.L. Trouilloud, A. Gupta, R.A. Wanner, W.J. Gal-
lagher, P.M. Rice, S.S.P. Parkin Appl. Phys. Lett. 77, 2373 (2000).
[4] B.D. Schrag, A. Anguelouch, G. Xiao, P.L. Trouilloud, Y. Lu, W.J. Gallagher, S.S.P. Parkin, J. Appl. Phys. 87,
4682 (2000).
[5] B.F.P. Roos, P.A. Beck, S.O. Demokritov, B. Hillebrands, J. Appl. Phys. 89, 6656 (2001).
[6] H.N. Yang, Y.P. Zhao, A. Chan, T.M. Lu, G.C. Wang, Phys. Rev. B 56, 4224 (1997).
AG Magnetismus
54
TU Kaiserslautern
6.8 High resolution magnetic patterning of exchange coupled multilayers
V.E. Demidov, D.I. Kholin, S.O. Demokritov, and B. Hillebrands1
The antiferromagnetic interlayer exchange coupling between two ferromagnetic layers separated
by a thin nonmagnetic spacer was discovered in 1986 by Grünberg et al. [1]. Since then this
phenomenon has been intensively studied (for a recent review see [2]) because of its potential
for the creation of antiferromagnetically coupled media for high-density magnetic recording [3, 4]
and artificial antiferromagnets that are used as hard magnetic electrodes for magnetic memory
elements [5, 6] and magnetoresistive sensors [7].
Recently it was shown [8], that the type of the interlayer exchange coupling between two iron
layers separated by a chromium spacer can be easily modified by ion irradiation. This modification
is a result of the atomic intermixing at the Fe/Cr interface caused by the dissipation of energy of
ions due to their collisions within the crystalline lattice. The intermixing at the Fe/Cr interface
leads to the appearance of microscopic “magnetic bridges” that connect the two iron films and
provide strong direct local ferromagnetic coupling between them. With increase of the irradiation
fluence the quantity of magnetic bridges increases and the type of interlayer coupling changes to the
ferromagnetic one. Since the direct coupling through a bridge is about hundred times stronger than
that via the spacer the density of bridges and, consequently, the ion fluences necessary to change
the type of coupling is very small [8]. It is obvious that the modification of the exchange interlayer
coupling by means of ion irradiation provides a very convenient tool for local modification of
magnetic properties of antiferromagnetically coupled layers. Due to the possibility to focus the
ion beam down to very few tens of nanometers the proposed technique opens the way for the
creation of novel artificial magnetic media such as thin-film structures containing nano-scaled
areas having different magnetic susceptibility. Moreover, due to the moderate ion penetration
depth into standard photoresist, which usually does not exceed 50 − 100 nm, one can realize ion
nano-patterning using standard resist mask technology.
In this paper we present first experimental results on the laterally resolved ion beam modification
of the interlayer exchange coupling in Fe/Cr/Fe trilayers.
For the experimental investigation samples were prepared consisting of two 10 nm thick Fe(001)
films separated by a 0.7 nm thick Cr spacer. The chosen thickness of the Cr spacer corresponds to
a strong antiferromagnetic coupling between the Fe layers. Using an ultra high vacuum molecular-
beam epitaxy system the Fe/Cr/Fe(001) trilayers were epitaxialy grown on a MgO (001) substrate
with a 80 nm thick Cr buffer and covered by 3 nm of Cr in order to avoid corrosion. Finally the
samples were irradiated using a focused ion beam lithography machine. Irradiation with a fluence
of 6.25 · 1015 ions/cm2 was performed with 50 keV Ga+ ions without applied external magnetic
field with the samples being kept at room temperature. The ion beam focused down to approxi-
mately 100 nm was scanned along the surface of the sample forming square shaped irradiated areas
with dimensions of 200 × 200 µm2 separated by 150 µm wide non-irradiated space [9]. The diago-
nals of the irradiated squares were aligned along the easy magnetic axes of the four-fold magnetic
anisotropy of the Fe(001) films.
The samples as prepared were studied using Magneto-Optical Kerr-Effect (MOKE) magnetometry.
The hysteresis loops were measured at different points of the samples. A magnetic field of up to
1 In collaboration with F. Wegelin and J. Marien, IBM Mainz Materials Laboratory, Mainz, Germany.
AG Magnetismus
55
TU Kaiserslautern
Fig. 1: Hysteresis loops for

1 a) 1 b) a non-irradiated (a) and an
Magnetization, a.u.
irradiated (b) area.
H
0 0
M1
M2
-1 HC1 HC2 -1
-1000 -500 0 500 1000 -1000 -500 0 500 1000

Applied field, Oe Applied field, Oe
1.5 kOe was applied parallel to the easy magnetic axis. Typical results of the MOKE-measurements
are presented in Fig. 1, where the two hysteresis loops shown correspond to a non-irradiated and
an irradiated area, respectively. In order to illustrate the rotation of magnetization in the sample
the vector diagram is added to Fig. 1a showing the relative orientation of the magnetic moments
M1 and M2 of the two Fe layers for different values of the external magnetic field H. As seen
from Fig. 1a, the non-irradiated trilayer structure exhibits typical antiferromagnetic magnetization
curves in the range of the external magnetic field of ±250 Oe. In this range the application of a
magnetic field leads to a weak deviation of the magnetic moments M1 and M2 from their original
antiparallel state, which results in the total magnetization of the trilayer to increase slowly with the
increase of the field strength. As soon as the strength of the magnetic field reaches the critical value
HC1 = 250 Oe, the total magnetization of the trilayer starts to increase stepwise. Such a behavior
can be explained by a weak 90◦ interlayer exchange coupling [10] coexisting in the samples with
the strong antiferromagnetic one. As a result, the orientation of the magnetic moments of the layers
almost at 90◦ with respect to each other becomes favorable in the range of magnetic fields from
250 to 450 Oe and M1 and M2 orient close to the directions of magnetic easy axes (see the vector
diagram in Fig. 1a). Finally, if the strength of the external magnetic field exceeds HC2 = 450 Oe
the trilayer becomes completely saturated. In contrast to the complicated magnetic behavior of
the non-irradiated Fe/Cr/Fe trilayer the irradiated trilayer simply exhibits a typical ferromagnetic
hysteresis loop (see Fig. 1b) with the coercive field equal to 20 Oe.
Next, Magnetic Force Microscopy (MFM) measurements were performed in the regions situated
close to the boundary between the irradiated and non-irradiated areas. For this purpose a multi-
functional scanning probe microscope was used (Solver AFM-MFM produced by NT-MDT Co.).
The microscope contains a built-in electromagnet, which allows one to investigate the domain
structure of samples placed in an external magnetic field.
Figure 2 presents the MFM images of the corner of the irradiated area obtained in weak external
magnetic fields, comparable to the coercive field of the irradiated area. The four panels a − d
of Fig. 2 correspond to the strength of applied magnetic field H equal to -30, 0, 5, and 30 Oe,
respectively. The magnetic field was applied along one of the magnetic easy axes, which are shown
in the figure by black arrows. As seen from Fig. 2a, in the magnetic field of −30 Oe the boundary
between the irradiated and non-irradiated areas provides very strong magnetic contrast, whereas the
remaining surface is magnetically uniform. Such a behavior is understood by the strong difference
in magnetizations of the irradiated and non-irradiated areas. The completely saturated irradiated
area produces strong magnetic stray fields at its edges if surrounded by the non-irradiated trilayer
with small net magnetization. As the external magnetic field decreases to zero (see Fig. 2b) the
AG Magnetismus
56
TU Kaiserslautern
Fig. 2: Magnetic images of a corner

of the irradiated area obtained in weak
external magnetic fields: −30 Oe (a),
0 Oe (b), 5 Oe (c), 30 Oe (d). The black
arrows indicate the easy axes of the
four-fold magnetic anisotropy of the
H=-30 Oe H= 0 Oe Fe(001) films.
H= 5 Oe H= 30 Oe
irradiated area breaks up into domains, whereas the non-irradiated area remains uniform. As the
magnetic field changes its direction and its strength increases again (see Fig. 2c and 2d) the domain
structure in the irradiated area changes and finally disappears in a field exceeding 20 Oe, whereas
the non-irradiated area does not demonstrate any visible change.
It is important to note here that for the whole range of magnetic fields between −30 and 30 Oe
the magnetic boundary between the irradiated and non-irradiated areas is seen very sharply. This
indicates that the strong magnetic moment of the irradiated ferromagnetic area does not have any
considerable influence on the magnetization of the non-irradiated antiferromagnetic area in the
boundary regions and, as a result, the change of magnetic properties at the boundaries between the
irradiated and non-irradiated areas occurs very abruptly. As the upper estimate for the length, on
which the change of magnetic properties occurs, the length of localization of the magnetic stray
field can be taken. From measurements a value of about 200 nm is estimated. Consequently, one
can conclude that ion irradiation provides a useful method for magnetic patterning of Fe/Cr/Fe
trilayers with a lateral resolution, which is in any case not worse than 200 nm.
Of particular interest is the change in the properties of the boundaries between the irradiated and
non-irradiated areas as a function of the external magnetic field. The MFM measurements show
that the magnetic boundaries remain well defined in magnetic fields of up to about ±200 Oe. As
the field strength approaches HC1 (see Fig. 1a) the magnetic images of the boundary regions start
to demonstrate qualitative changes. This fact is illustrated by Fig. 3 where the MFM images are
shown measured for the strength of the magnetic field equal to 210 Oe (a) and 250 Oe (b).
Unfortunately, in strong external magnetic fields the domain structure of the sample cannot un-
ambiguously be derived from the MFM measurements due to the rotation of magnetization of the
MFM tip out of its axis. However, it is clearly seen from Fig. 3 that in the boundary region the
magnetization of the non-irradiated trilayer experiences a strong influence of the irradiated area.
This phenomenon is caused by an instability of the magnetization in an antiferromagnetically cou-
pled non-irradiated trilayer in magnetic fields lying close to HC1 . Due to the presence of the 90◦
coupling the alignment of the magnetic moments of the two Fe layers at 90◦ becomes favorable in
the range of magnetic field between HC1 and HC2 . The strong magnetic stray field of the irradiated
area stimulates the nucleation of the 90◦ phase in the non-irradiated area close to the boundary. As
AG Magnetismus
57
TU Kaiserslautern
Fig. 3: Magnetic images of a corner of

the irradiated area obtained in strong
external magnetic fields: 210 Oe (a),
250 Oe (b).
H= 210 Oe H= 250 Oe
a result the magnetic boundary between the irradiated and non-irradiated areas becomes wide.
In conclusion, we have demonstrated that local ion beam modification of the antiferromagnetic in-
terlayer exchange coupling in Fe/Cr/Fe trilayers can be successfully used for magnetic patterning
on the submicrometer scale. Such patterning provides a unique mean to create antiferromagnetic
thin-film structures containing micrometer-size ferromagnetic areas with a very abrupt change of
magnetic properties at their boundaries. The performed investigation shows that the well-defined
magnetic boundary between irradiated and non-irradiated areas exists in a large range of the exter-
nal magnetic field.
A full report has been submitted to Applied Physics Letters [11].
The work is supported in part by the European Communities Human Potential Programme (con-
tract number HPRN-CT-2002-00296 NEXBIAS) and by the Deutsche Forschungsgemeinschaft
through the Priority Programme 1133 “Ultrafast Magnetization Processes”.
References
[1] P. Grünberg, R. Schreiber, Y. Pang, M.B. Brodsky, H. Sowers, Phys. Rev. Lett. 57, 2442 (1986).
[2] D.E. Bürgler, S.O. Demokritov, P. Grünberg, M.T. Johnson, Handbook of Magnetic Materials, vol. 13,
Ed. K.J. H. Buschow, Elsevier, Amsterdam (2001).
[3] E.E. Fullerton, D.T. Margulies, M.E. Schabes, M. Carey, B. Gurney, A. Moser, M. Best, G. Zeltzer, K. Rubin,
H. Rosen, M. Doerner, Appl. Phys. Lett. 77, 3806 (2000).
[4] K. Tang, M. Doerner, Q.-F. Xiao, L. Tang, M. Mercado, J. He, R. Prichard, P. Rice, J. Appl. Phys. 93, 7402
(2003).
[5] J. Schmalhorst, H. Brueckl, G. Reiss, R. Kinder, G. Gieres, J. Wecker, Appl. Phys. Lett. 77, 3456 (2000).
[6] H. Brueckl, J. Schmalhorst, H. Boeve, G. Gieres, J. Wecker, J Appl. Phys. 91, 7029 (2002).
[7] K. Bal, H.A.M. van den Berg, D. Deck, Th. Rasing, J Appl. Phys. 90, 5228 (2001).
[8] S.O. Demokritov, C. Bayer, S. Poppe, M. Rickart, J. Fassbender, B. Hillebrands, D.I. Kholin, N.M. Kreines,
O.M. Liedke, Phys. Rev. Lett. 90, 97201-1 (2003).
[9] The relatively large size of the irradiated squares was chosen in order to allow local magnetooptical measure-
ments of hysteresis loops in different areas.
[10] S.O. Demokritov, J. Phys. D 31, 925 (1998).
[11] V.E. Demodov, D.I. Kholin, S.O. Demokritov, B. Hillebrands, submitted to Appl. Phys. Lett.
AG Magnetismus
58
TU Kaiserslautern
6.9 Measurement of the three magnetization vector components of a Ni81 Fe19

film
H. Nembach, M.C. Weber, J. Fassbender, and B. Hillebrands
The magnetization dynamics in thin magnetic films can be investigated in the frequency domain
with Brillouin light scattering (BLS) and ferromagnetic resonance (FMR), and in the time domain
with inductive, resistive and optic techniques. One main advantage of the optic techniques in the
time domain is that they allow one to measure all three components of the magnetization vector.
This is important for studies on precessional dynamics where large-angle excitations are consid-
ered. The knowledge of the experimentally determined trajectory allows for a comparison with
numerical simulations. The importance of spin wave propagation effects for the damping process
can also be evaluated. If, for example, the magnitude of the magnetization vector is conserved, the
generation of spin waves is an unlikely contribution to the damping mechanism.
The calibrated method for the determination of all three magnetization vector components in
time and space presented here goes back to Ding et al. [1], who employed this procedure for the
measurement of the spin reorientation transition in Co/Au(111). For the measurements a typical
magneto-optic Kerr effect setup in longitudinal geometry is used. The light is s-polarized and the
rotation of the polarization upon reflection at the sample surface is given by:
θKerr = θL · mL + θP · mP (1)
with θKerr the measured Kerr rotation, θL the longitudinal magneto-optical constant, θP the polar
magneto-optical constant, mL the longitudinal and mP the polar magnetization component. The
polarization of the incident light is not sensitive to the transversal magnetization component. The
knowledge of the two magneto-optical constants θL and θP is essential to determine the magne-
tization vector components. In order to determine these two constants an in-plane and an out-of
plane hysteresis curve is measured. The following procedure is employed to measure all three
magnetization components, see Fig. 1.
The static and the dynamic magnetic field are both in the film plane and oriented perpendicular to
each other. The frame of reference is fixed with the sample. After each measurement the sample
together with the dynamic and static magnetic field is rotated counterclockwise by 90◦ . The three
Fig. 1: The three different orientation of the setup which are used to determine all three components of the
magnetization vector [2].
AG Magnetismus
59
TU Kaiserslautern
Fig. 2: a) The time evolution of the three calibrated and normalized magnetization components mx , my and
mz are shown by open and filled circles and by squares, respectively. The lines show the results of a simula-
tion based on the Landau-Lifshitz-Gilbert equation assuming a pulse with linear slopes. b) Trajectory of the
magnetization vector after application of the magnetic field pulse. The equilibrium position is [1, 0, 0]. The
termination of the field pulse is indicated by the black dot. The arrow marks the direction of the applied field
pulse.
magnetization components mx , my and mz can then be determined in the following way:
1
mz = · (θKerr (0◦ ) + θKerr (180◦ )) (2-a)
2θP
1
my = · (θKerr (0◦ ) − θKerr (180◦ )) (2-b)
2θL
1
mx = · (θKerr (90◦ ) − θP mz ) (2-c)
θL
θKerr (0◦ ), θKerr (90◦ ) and θKerr (180◦ ) are the measured Kerr rotation angles in the respective con-
figurations.
In order to demonstrate this procedure the magnetization trajectory is determined for a 20 nm
thick Ni81 Fe19 film. The film has been sputter deposited in an Edwards ESM 100 system with a
base pressure of 3.6 · 10−6 mbar on a 10 × 10 mm2 MgO substrate. It is (001)-oriented and has a
thickness of 100 µm. The static bias field and the dynamic field pulse are 1440 A/m (18 Oe) and
800 A/m (10 Oe), respectively. The pulse length is 750 ps and its rise and fall times are estimated
to be 600 ps and 900 ps respectively. The estimations are based on measurements with a high
bandwidth oscilloscope on an un-shortened stripline with similar dimensions. These rather large
rise and fall times are due to the impedance mismatch of the stripline with the connecting cables.
The magneto-optical constants for the Ni81 Fe19 film are determined to be θL =0.038 mdeg/G and
θP =0.36 mdeg/G. The time-resolved measurements of the magnetization dynamics have been
performed in all three measurement geometries. The time evolution of the magnetization compo-
nents mx (open circles), my (close circles) and mz (squares) is shown in Fig. 2a.
The precessional frequency ν =1.4 GHz and the damping constant α = 0.008 are determined from
a comparison to a simulation based on the Landau-Lifshitz-Gilbert equation (solid line). The
AG Magnetismus
60
TU Kaiserslautern
Fig. 3: Temporal evolution of the square of the magnetization vector. The magnitude of the magnetization
vector is reduced by about 10 %. The dashed line is the result from the Landau-Lifshitz-Gilbert equation in
macrospin approximation, with the magnetization vector magnitude conserved.
magnetization trajectory is plotted in Fig. 2b. The equilibrium position of the magnetization vector
is [1, 0, 0] and the field pulse is applied in the [0, 1, 0] direction. The termination of the field pulse
is indicated with the black dot.
Figure 3 shows the temporal evolution of the square of the magnetization. A reduction of 10 %
is observed. This indicates the generation of spin waves. These spin waves can be either directly
excited by the stripline or by multi-magnon processes. Due to our limited lateral resolution we
cannot distinguish between these two origins.
In summary, a method for the determination of the magnetization trajectory is demonstrated. A
reduction of the magnetization magnitude is observed. This is attributed to the generation of spin
waves.
We like to thank R. Lopusnik for stimulating and helpful discussions and M. Thornton for provid-
ing the Ni81 Fe19 sample.
A full report has been submitted to Journal of Applied Physics [3].
References
[1] H. Ding, S. Pütter, H. Oepen, J. Kirschner Phys. Rev. B 63, 134425 (2001).
[2] R. Lopusnik, Dissertation, Universität Kaiserslautern (2001).
[3] H. Nembach, M. Weber, J. Fassbender, B. Hillebrands, submitted to J. Appl. Phys.
AG Magnetismus
61
TU Kaiserslautern
D. Exchange Bias Effect
6.10 Optical control of the magnetization in exchange biased NiFe/FeMn

bilayers on the picosecond timescale
M.C. Weber, H. Nembach, J. Fassbender
Active research is focussed towards a deeper understanding of the origin of unidirectional aniso-
tropy in bilayer systems consisting of a ferromagnetic (F) and an adjacent antiferromagnetic (AF)
layer, both for fundamental and technological reasons. This unidirectional anisotropy is found for
instance if the F-AF-bilayer system is cooled below the blocking temperature of the AF layer in an
applied magnetic field. It is of large interest to investigate the exchange bias effect on short time
scales, in particular in view of fast memory applications such as magnetic random access memory
(MRAM) and media for hard disk and magneto-optical data storage [1–3].
Considerable insight has been gained from quasi-static magnetization reversal studies and theo-
retical micromagnetic modelling, indicating that the microstructure of the hetero-interface plays a
key role in such systems [4–6]. For the dynamical properties two major effects must be consid-
ered here: First, using ultrashort laser pulses with durations in the range of several femtoseconds,
the interplay between the spin-system and the electronic and phononic systems is time dependent.
This has been studied by Ju et al. [7] who examined the NiFe/NiO exchange bias system by means
of all-optical femtosecond pump-probe experiments. Second, and studied here, thermal heating of
the interface is achived using picosecond laser pulses, and the temporal evolution is largely gov-
erned by the dynamics of heat flow. In both cases, precessional effects will also contribute, if the
length of the laser pulse is shorter than the characteristic precession time.
The polycrystalline exchange bias samples have been prepared by UHV e-beam evaporation. As
a growth template a 15 nm thick Cu buffer layer on top of a thermally oxidized Si substrate has
been used. The exchange bias system itself consists of a 5 nm thick Ni81 Fe19 (F) layer and a 10 nm
thick Fe50 Mn50 (AF) layer. Finally the sample is covered by a 2 nm Cr cap layer to protect it from
oxidation. In order to initialize the exchange bias effect the sample was field cooled from above
the blocking temperature (155 ◦ C) after deposition. For further details see [8].
The time resolved experiments have been performed using a pump-probe technique in the follow-
ing way (see also Chapter 6.17). A 9 ps short laser pulse is generated by a mode-locked diode
pumped Nd:YVO4 laser oscillator running with a repetition rate of 80 MHz and a wavelength of
1064 nm. The laser beam is inserted into a second-harmonic unit which maintains the pulse dura-
tion and delivers laser pulses of 532 nm wavelength. The linear polarization of the outgoing beam
is defined by a combination of a retarding wave plate and a Glan-Laser polarizer. The polarized
beam is then divided into an intense pump and a weak probe laser pulse by a beamsplitter. The
7 nJ pump pulse is directed nearly normal to the sample surface and focused to a spot diameter of
25 µm. The probe pulse is time delayed by a translation stage covering a time interval from a nega-
tive delay of −400 ps up to about 6 ns. The probe beam is used to sense the pump induced changes
of the magnetization of the NiFe layer by means of the longitudinal magneto-optic Kerr effect (spot
diameter: 20 µm). The sample has been patterned into 100 × 100 µm2 squares by means of optical
lithography and wet chemical etching. The pump and probe spots are adjusted to the center of a
square by maximizing the observed reflected light intensity and reducing stray reflections. Thus
AG Magnetismus
62
TU Kaiserslautern
2
Fig. 1: Easy axis magnetization reversal
0 - 370 ps loop for three different pump-probe delay
-2 times as indicated in the figure. The ini-
-4 Heb tial exchange bias field Heb is indicated by a
-6 black arrow and the dashed lines are guides
-8
to the eye.
(mdeg)
-10
-12
2
0
+ 20 ps
K
-2
Kerr rotation
-4
-6
-8
-10
-12
2
0
+ 125 ps
-2
-4
-6
-8
-10
-12
0 25 50 75 100 125 150 175 200 225 250
magnetic field H (Oe)
a good spatial overlap of both beams was achieved. In order to investigate the optical control of
the magnetization in the time domain a quasi-static hysteresis loop is sensed by a probe pulse with
a fixed time delay to the laser excitation pulse. The quasi-static hysteresis loop then reflects the
magnetic parameters present for a given time delay.
First, the quasi-static hysteresis loop along the easy magnetization direction is investigated (see
top graph in Fig. 1). This measurement has been performed using a negative time delay of 370 ps
between pump and probe pulse, i. e., the probe pulse arrives 12.13 ns after the pump pulse. Within
this time range all time-dependent effects are relaxed already (see Fig. 2). An exchange bias field
of Heb = 123 Oe and a coercive field of HC = 24 Oe are observed. For a positive delay time of
20 ps a maximum reduction of the exchange bias field to a value Heb = 67 Oe is observed. The
coercive field is also reduced correspondingly. For a time delay of 125 ps a partial recovery of the
exchange bias field is already observed. In order to analyze in more detail the hysteresis loops
with a fixed time delay to the pump pulse the exchange bias field values for time delays in the
range between −400 ps and 3000 ps are plotted as a function of delay time in Fig. 2. Within the
first 20 ps a fast reduction of the exchange bias field of about 45 percent can clearly be observed,
followed by a slower recovery to the initial bias field value. The open dots in Fig. 2 represent the
exchange bias field values extracted from the three loops presented in Fig. 1. The data have been
fitted to a phenomenological model introduced by Ju et al. [7]. This model describes the process
of photomodulation by a time dependent single exponential driving term,
Heb = Heb,init (1 − m · exp (−t/τ)) (1)
where Heb,init describes the initial bias field value, m is the so called modulation depth, e.g., the
strength of the photomodulation and τ represents the time constant of the recovery of the exchange
AG Magnetismus
63
TU Kaiserslautern
exchange bias field Heb (Oe) 130 Fig. 2: Time dependance of the ex-
120 change bias field Heb (t) of an easy
axis magnetization reversal loop. The
110 open dots correspond to the hysteresis
100
curves shown in Fig. 1. A fit of Eq. (1)
with the parameters m = 0.45 and τ =
90 170 ps is shown by a dashed line.
80
70
60
-500 0 500 1000 1500 2000 2500 3000
delay time (ps)
bias shift field. The best fit to our experimental values is shown as a dashed line in Fig. 2 with
m = 0.45 and τ = 170 ps.
Next the zero field susceptibility of the hard axis magnetization reversal loops is investigated. It
is addressed, whether the time dependence matches the one of the exchange coupling modulation
found in the easy axis case. The quasi-static magnetization reversal loops taken at a negative
delay time (not shown) show the expected hard axis behavior for exchange bias systems. Shortly
after the arrival of the excitation pulse a dramatic change in the hard axis loop shape is observed.
20 ps after the heating pulse the loop shape resembles the shape of a NiFe easy axis loop. For
larger delay times the hard axis behavior is starting to be restored. In order to analyze these data
more quantitatively the zero field susceptibility has been extracted from the magnetization reversal
curves by performing a linear fit to the data close to zero applied field. The slope corresponds to the
zero field susceptibility χ, which is plotted in Fig. 3 as a function of the pump-probe delay time.
Within the first 20 ps a sharp increase of χ is observed indicating a transition of hard magnetic
magnetization reversal behavior to a nearly pure easy axis magnetization reversal hysteresis curve
of NiFe. For larger delay times the susceptibility again reaches its initial value. Again the time
constant involved is extracted by a fit to Eq. (1) with a modulation depth m∗ and a recovery time
τ∗ . The best fit is achieved with the values of τ∗ = 160 ps and m∗ = 13.75 ps (dashed line in Fig. 3).
The time constants for easy and hard axis magnetization reversal are in good agreement indicating
that both time dependencies rely on the mechanism of fast unpinning and recovery of the interfacial
exchange coupling.
The experimental observation of both, the fast reduction of the exchange bias field and the strong
increase of the susceptibility on a 20 ps timescale, can be understood in terms of a fast thermal
unpinning of the exchange coupling at the F/AF hetero-interface upon arrival of the pump laser
pulse. Within the laser pulse duration the temperature of the lattice system is elevated close to the
blocking temperature of the exchange bias system. The exchange coupling is thus reduced and a
nearly isotropic easy axis magnetization reversal behavior is observed. The relaxation time for the
slow recovery process of both the exchange bias field and the susceptibility can be understood in
terms of energy dissipation. The lattice system is cooled by heat flow from the metallic bilayer
to the substrate or to a region outside the laser spot, setting the ultimate limit for the speed of
recovery.
Currently the reduction of the exchange bias field is limited to about 45 percent of the initial value
because for larger pump pulse energies irreversible processes start to occur, which obstruct the
stroboscopic measurement scheme used here. Since the samples are polycrystalline, a distribution
AG Magnetismus
64
TU Kaiserslautern
0,7 Fig. 3: Time dependance of the zero

susceptibility (mdeg/Oe) 0,6 field susceptibility χ(t) of a hard axis
magnetization reversal loop. The
0,5 dashed line represents a fit to a mod-
0,4 ified version of Eq. 1 with a time con-
stant of τ∗ = 160 ps.
0,3
0,2
0,1
0,0
-250 0 250 500 750 1000 1250 1500 1750 2000
delay time (ps)
of different AF grain sizes is present. According to Takano et al. [9] large AF grains at the interface
of an exchange bias system exhibit a smaller exchange coupling to the F spins across the interface.
Upon laser excitation these grains might already switch completely and will not relax to the orig-
inal magnetization direction. In order to minimize the distribution width of the exchange-bias
field values, measurements of epitaxial exchange bias systems are planned. Due to the different
magnetization reversal processes in these epitaxial systems for easy and hard axis magnetization
reversal also different time constants are expected. From the applications point of view it should be
possible to optically trigger a coherent magnetization rotation which for instance can be used for
fast thermo-magnetic writing and precessional dynamics of the F layer in an exchange bias layer
system on a picosecond time scale.
This research was actively supported by Lumera Lasers GmbH in setting up and developing the
laser equipment (see Chapter 6.17). The authors would like to thank S. Poppe for the sample prepa-
ration, H. Fouckhardt for the optical lithography support and B. Hillebrands for valuable discus-
sions. M. Weber would like to acknowledge support by the Graduiertenkolleg “Nichtlineare Optik
und Ultrakurzzeitphysik” of the DFG. The work is supported in part by the European Communities
Human Potential programme under contract number HRPN-CT-2002-00318 ULTRASWITCH.
A full report has been submitted to Journal of Applied Physics [10].
References
[1] J. Nogués, I.K. Schuller, J. Magn. Magn. Mater. 192, 203 (1999).
[2] R.L. Stamps, J. Phys. D: Appl. Phys. 33, R247 (2000).
[3] A.E. Berkowitz, K. Takano, J. Magn. Magn. Mater. 200, 555 (1999).
[4] A.P. Malozemoff, Phys. Rev. B 37, 7673 (1988).
[5] T.C. Schulthess, W.H Butler, J. Appl. Phys. 85, 5510 (1999).
[6] H. Ohldag, A. Scholl, F. Nolting, E. Arenholz, S. Maat, A.T. Young, M. Carey, J. Stöhr, Phys. Rev. Lett. 91,
017203-1 (2003).
[7] G. Ju, A.V. Nurmikko, R.F.C Farrow, R.F. Marks, M.J. Carey, B.A. Gurney, Phys. Rev. Lett. 82, 3705 (1999).
[8] A. Mougin, T. Mewes, M. Jung, D. Engel, A. Ehresmann, H. Schoranzer, J. Fassbender, B. Hillebrands, Phys.
Rev. B 63, 060409(R) (2001).
[9] K. Takano, R.H. Kodama, A.E. Berkowitz, W. Cao, G. Thomas, J. Appl. Phys. 83, 6888 (1998).
[10] M.C. Weber, H. Nembach, J. Fassbender, submitted to J. Appl. Phys.
AG Magnetismus
65
TU Kaiserslautern
6.11 Investigation of the polycrystalline Fe19 Ni81 /Fe50 Mn50 exchange bias
system with varying Cu spacer layer for partial decoupling
M.O. Liedke, H. Nembach, J. Fassbender, and B. Hillebrands
The exchange coupling across an interface between a ferromagnetic and an antiferromagnetic layer
may result in the so-called exchange bias effect evidenced by a shift of the hysteresis loop along
the magnetic field axis. To study the role of the exchange interaction at and near the interface,
Fe19 Ni81 /Fe50 Mn50 bilayers have been studied, which have an intervening Cu layer of varying
thickness and position in the antiferromagnetic Fe50 Mn50 layer. The role of the intervening Cu
layer is to generate partial exchange decoupling.
We have prepared and investigated a polycrystalline Fe19 Ni81 /Fe50 Mn50 bilayer with an interven-
ing wedge-shaped (0 − 0.7 nm) Cu layer. As the growth template a Si(111) wafer with a 15 nm
thick Cu buffer layer was used. The thickness of the ferromagnetic Fe19 Ni81 layer is 5 nm. On
the top a wedge (0 − 10 nm) of the antiferromagnetic material Fe50 Mn50 was grown. Next, the
Cu spacer wedge was grown with the wedge direction perpendicular to the wedge direction of the
Fe50 Mn50 film. Finally, a wedge of Fe50 Mn50 was grown on top with the opposite wedge direc-
tion compared to the first Fe50 Mn50 layer resulting in a constant sum of the thicknesses of both
Fe50 Mn50 layers (see the sketches in Fig. 1). For protection a 2 nm thick Cr layer was deposited.
As a result, a sample was obtained, where in one direction the position of the intervening Cu layer
varies from the interface to the top surface of the Fe50 Mn50 layer at constant Cu thickness, and in
the other direction the Cu thickness varies. To initialize the exchange bias effect the sample was
annealed above the Néel temperature and cooled in an applied field back to room temperature.
Fig. 1: Two dimensional scans of the sample: (a) exchange bias field Heb and (b) coercive field Hc as function
of the Cu spacer layer tCu (vertical direction) and the thickness tFeMn of the lower Fe50 Mn50 layer, which is
directly exchange coupled to the Fe19 Ni81 layer (horizontal direction).
AG Magnetismus
66
TU Kaiserslautern
Fig. 2: (a) Exchange bias field Heb and (b) coercive field Hc as a function of the Cu spacer layer tCu for different
thicknesses tFeMn of the Fe50 Mn50 layer coupled with FeNi.
Hysteresis loops were measured as a function of position using magneto-optic Kerr effect magne-
tometry. From the loops the exchange bias field Heb and the coercive field Hc were extracted as
a function of the Cu spacer thickness, tCu , and the thickness tFeMn of the lower Fe50 Mn50 layer,
which is directly exchange coupled to the Fe19 Ni81 layer. The results are shown in Figs. 1, 2 and 3.
The obtained two-dimensional maps provide us easily with global information about the influence
of the partial decoupling of the antiferromagnet as a function of tCu and tFeMn . To obtain more
quantitative information several line scans along the horizontal and vertical axes in Fig. 1 were
evaluated. With increasing Cu thickness the absolute value of the exchange bias field Heb decreases
exponentially, see Fig. 2a. The coercive field Hc decreases rapidly for small values of tFeMn , and
it stays stable for tFeMn > 3 nm achieving a maximum value of about 50 Oe, see Fig. 2. In case
of tFeMn = 0 the maximum value of the exchange bias field was not obtained for tCu approaching
zero. This is likely due to the interrupted growth necessary for creating the intervening Cu layer.
However a line scan along the Cu wedge for tFeMn = 0 yields a dependence in good agreement
with previous results [1].
Fig. 3: (a) Exchange bias field Heb and (b) coercive field Hc as a function of the thickness tFeMn of the lower
Fe50 Mn50 layer coupled to FeNi for different thicknesses tCu of the Cu spacer layer.
AG Magnetismus
67
TU Kaiserslautern
In Fig. 3 the strong influence of the Cu spacer layer on the exchange bias field Heb and the coercive
field Hc as a function of its position in the stack, tFeMn , is shown. Even for a 0.1 nm thick Cu
spacer layer we observe a strong decrease of exchange bias value, which corresponds to a decrease
of interfacial torques due to the partial decoupling, see Fig. 3a. On the other hand an influence of
the upper wedge of Fe50 Mn50 is observed, which is strong enough to maintain the unidirectional
anisotropy. Considering Fig. 3b the coercive field exhibits a maximum for about tFeMn = 4 nm. As
a function of the Cu spacer thickness, tCu , the width of the maximum varies. However the position
of the maximum in the coercive field does not follow the exchange bias field (Fig 3a) and stays
stable for the same value of tFeMn . This indicates that the coercive field is not directly connected
to the origin of exchange bias coupling.
References
[1] T. Mewes, B. F. P. Roos, S. O. Demokritov, B. Hillebrands Appl. Phys. 87, 5064 (2000).
AG Magnetismus
68
TU Kaiserslautern
6.12 Structural analysis of the Ni81 Fe19 /Fe50 Mn50 exchange bias layer sys-
tem
S. Blomeier, J. Fassbender, and B. Hillebrands1
The results presented here have been obtained during and following a stay of Steffen Blomeier
in the group of Prof. J. Chapman at the University of Glasgow. For the structural analysis of the
investigated layer system transmission electron microscopy studies were carried out there.
Recently it has been shown that the exchange bias effect in the polycrystalline Ni81 Fe19 /Fe50 Mn50
exchange bias system, as well as in similar systems, can be modified by means of He+ -ion irra-
diation [1, 2], as shown in Fig. 1. Subject of this report is a careful search for irradiation induced
structural changes using transmission electron microscopy (TEM).
If the irradiation is carried out without the application of an external field, an enhancement of the
bias field for doses of up to 4 · 1015 ions/cm2 is found, whereas for larger doses a decrease of the
bias field occurs. If the irradiation is carried out in a field, which is antiparallel to the field Hprep ,
applied during the field cooling procedure of the investigated sample, a change of the sign of the
bias field can occur. In addition, an enhancement of the absolute value of the bias field is also
found in this case, albeit the effect occurs at slightly higher doses and is less pronounced than in
the case of a zero external field. For higher doses, a suppression of the bias field value is also
found here. A model was proposed to understand the observed phenomena by attributing them to
the creation of point defects [1, 3].
Fig. 1: Dependence of the exchange bias field of a polycrystalline Ni81 Fe19 /Fe50 Mn50 system on the ion
dose for an irradiation with 5 keV He+ -ions. The field axis is normalized to the initial bias field value for a
non-irradiated sample.
1 In
collaboration with D. McGrouther, S. McVitie and J. Chapman, Department of Physics and Astronomy,
University of Glasgow, Glasgow, United Kingdom.
AG Magnetismus
69
TU Kaiserslautern
Fig. 2: Layer sequence of the investigated sample.
The structural analysis of a series of samples was carried out using the JEOL 2000FX electron
microscope in Glasgow. To ensure the comparability of the obtained results, all samples were
grown simultaneously in a UHV MBE system and subsequently irradiated with different doses
of 5 keV He+ ions. For the TEM studies samples are required to be electron-transparent, and
they were thus grown on special window substrates, as shown in Fig. 2. These substrates consist
of a 50 nm thick membrane of amorphous Si3 N4 , on top of which the layer system is deposited.
The investigated layer system itself consisted of a 15 nm thick Cu buffer, followed by 10 nm of
antiferromagnetic Fe50 Mn50 , 5 nm of ferromagnetic Ni81 Fe19 , and a chromium cap layer of 2 nm
thickness.
Four samples irradiated with doses of 0, 2 · 1015 , 1 · 1016 , and 1 · 1017 ions/cm2 were investigated.
Bright field images, dark field images and diffraction patterns were taken from each sample. The
bright field and dark field images were used to determine the particle size distribution in the sam-
ple layers, while the diffraction patterns were used to determine their crystallographic structure,
including lattice type, lattice constant and degree of texture. Figure 3 shows a typical result.
For the analysis of the bright field images a program written by T. Mewes was used, which allows
one to determine the correlation length in a given greyscale image, which is taken as a measure for
Fig. 3: (a): Bright field image of a sample irradiated with a dose of 1 · 1017 ions/cm2 . (b): Dark field image
from the same sample, on the same spot. (c): Diffraction pattern from the same sample.
AG Magnetismus
70
TU Kaiserslautern
40
30
Counts
20
10
0
0 200 400
Particle size [nm²]
Fig. 4: Particle size distribution from the sample shown in Fig. 3. Particles are required to have an area of at
least 20 nm2 , to exclude the effects of noise.
the average particle size [4]. For the analysis of the dark field images the software package ImageJ
was used. A particle size distribution, as shown in Fig. 4, can be obtained directly by this method.
From this size distribution, an average particle size can be derived, by taking the square root of the
mean particle area as a measure for the average particle size.
An average particle size of 10.5 ± 2 nm is found. This size does not change within error bars for
an increasing ion dose.
For the analysis of the diffraction patterns, a “trial-and-error”-method has been employed. This
method starts with a reasonable guess of the crystallographic structure of the investigated sample.
In this case, it was assumed that the sample has an fcc structure, because Cu, Fe50 Mn50 , and
Ni81 Fe19 all grow fcc as bulk materials with only slightly different lattice constants.
2.2
Experiment
2.0 Calculation
1.8
1.6
relative intensity [a.u.]
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-10 0 10 20 30 40 50 60 70
radial position [mm]
Fig. 5: Comparison between the calculation and the experimental data. The suspected fcc structure is fairly
well reproduced.
AG Magnetismus
71
TU Kaiserslautern
A result is depicted in Fig. 5. The sample shows a fcc structure. The positions of the outermost
peaks are slightly shifted, and these peaks are also slightly broadened and thus reduced in intensity.
This can be explained as a result of the fact that there is no uniform lattice constant present within
the investigated layer system but a distribution of lattice constants which are all of similar value.
Moreover, the innermost peak is not well reproduced due to saturation effects of the recording film.
No experimental peak exceeds its theoretically predicted value, which means that there is no texture
present in the investigated sample. This can be confirmed by tilting the sample within the electron
microscope setup and thus scanning all possible spatial directions. It was found that the diffraction
images do not change their properties with increasing tilt angles.
As the most important result it was found that irradiation with He+ ions does not change the
structure for the investigated dose regime within the sensitivity of the characterization.
References
[1] S. Poppe, Modifikation des Austausch-Verschiebungseffektes in Ferro-/Antiferromagnet-Bilagen durch lagen-
weise und ortsaufgelöste Ionenbestrahlung, Dissertation, TU Kaiserslautern (2003).
[2] A. Mougin, T. Mewes, M. Jung, D. Engel, A. Ehresmann, H. Schmoranzer, J. Fassbender, B. Hillebrands Phys.
Rev. B 63, 060409(R) (2001).
[3] J. Juraszek, J. Fassbender, S. Poppe, T. Mewes, B. Hillebrands, D. Engel, A. Kronenberger, A. Ehresmann, H.
Schmoranzer J. Appl. Phys. 91, 6896 (2002).
[4] T. Mewes, Systematik epitaktischer Austausch-Verschiebungsschichtsysteme, Dissertation, TU Kaiserslautern
(2002).
AG Magnetismus
72
TU Kaiserslautern
6.13 Modification of the magnetic properties of the polycrystalline

Ni81 Fe19 /Fe50 Mn50 exchange bias bilayer system by Ga+ -ion irradia-
tion
S. Blomeier, J. Fassbender, and B. Hillebrands1
The results presented here have been obtained during and following a stay of Steffen Blomeier
in the group of Prof. J. Chapman at the University of Glasgow. For the analysis of the magnetic
properties of the investigated layer system Lorentz microscopy studies were carried out there. Also
access was provided to the local focused ion beam machine for Ga ion irradiation.
Recently it has been shown that the exchange bias effect in the polycrystalline Ni81 Fe19 /Fe50 Mn50
exchange bias system, as well as in similar systems, can be modified by means of He+ -ion irradi-
ation, as discussed in detail in [1, 2]. Both an enhancement of the exchange bias field as well as a
reversal of the bias direction has been found. It is of large interest, whether such behavior can also
be achieved by using ions of larger mass and kinetic energy. Subject of this study are irradiation
experiments using 30 keV Ga+ -ions provided by a focused ion beam machine.
For 5 − 10 keV He+ -ions the obtained dose dependence is well simulated by a model function [1,2]
Hbias (N)
= (1 ± aptN)(exp(−bN)) , (1)
Hbias,initial
which attributes the observed phenomena to the creation of point defects in different regions of the
layer system. Here, t describes the thickness of the antiferromagnet, p is an efficiency constant,
describing the number of created defects per ion, N is the dose of the ions applied during irradi-
ation, while a and b denote two efficiency parameters for the influence of such point defects on
the bias field value. The first factor describes the observed enhancement effects, while the second
factor models the decrease of the bias field. While the enhancement effect is attributed to the cre-
ation of point defects within the volume of the antiferromagnet, the decrease of the bias field is
believed to be a result of the creation of defects at the interface between the ferromagnet and the
antiferromagnet. The “+”(“-”) sign refers to parallel (antiparallel) alignment between the magnetic
fields for initialization of the exchange bias effect and the field applied during ion irradiation.
Results on a different system, a sputtered FeCo/IrMn double layer irradiated by 30 keV Ga+ -ions,
have been obtained by D. McGrouther at the University of Glasgow. Contrary to our previous
results on NiFe/FeMn, he only found a decrease of the bias field strength.
The irradiation was carried out using an FEI Strata 200XP Focused Ion Beam system as a source
for 30 keV Ga+ -ions. For this purpose, a special dose test pattern designed by D. McGrouther was
applied to a sample which was then subsequently investigated by means of Lorentz microscopy.
For this analysis, a Philips CM20 electron microscope with a modified objective lens was used in
the Fresnel mode. For this purpose, a special electron-transparent window substrate was used and
the deposition of the exchange bias system was carried out in a UHV MBE system (see Chapter
6.12).
1 In
collaboration with D. McGrouther, S. McVitie and J. Chapman, Department of Physics and Astronomy,
University of Glasgow, Glasgow, United Kingdom.
AG Magnetismus
73
TU Kaiserslautern
76 Oe 252 Oe 302 Oe 352 Oe
Fig. 1: Fresnel images from the non-irradiated rectangle of the dose test pattern at different values of the applied
external field, showing a magnetic reversal process. The applied field is indicated.
The dose test pattern mentioned above consists of nine adjacent rectangles that are 10 × 400 µm2
in size. Each rectangular region was irradiated with a different ion dose in the range (0 − 5) ·
1015 ions/cm2 . The rectangles are separated by thin lines which were irradiated by a dose of
1.25 · 1016 ions/cm2 .
Fresnel images were taken from each of those rectangles that still exhibited a magnetic contrast
after irradiation. For each such region, a complete hysteresis loop in steps of 25 − 50 Oe was
recorded. Figure 1 shows a series of these Fresnel images from the non-irradiated region of the
dose test pattern at different values of the applied external field.
le f t right
From these images, values for Hc and Hc , the left and right coercive fields of each loop, were
extracted, and from these values the respective bias fields and coercivity values were calculated.
Figure 2 shows the dependence of the bias field and the coercivity of the ion dose, respectively. No
enhancement for the bias-field is found for this layer system. The bias field value instead decays
exponentially, as modelled by the second factor in Eq. (1).
An exponential fit of the data for the bias field dependency yielded a parameter value of b =
6.0299 · 10−15 cm2 , while the parameter a was set to a value of zero. This value was used to plot
the same exponential decay function into the diagram of the experimental data for the coercivity.
1,6
1,2
Experimental Data 1,4 Experimental Data
Exponential Fit Exponential Fit
1,0 1,2
0,8 Model: exp(-b*N) 1,0

Model: exp(-b*N)
0,8
b= 6.0299E-15
HEB/HEB,initial
HEB/HEB, initial
0,6 b= 6.0299E-15
0,6
0,4 0,4
0,2
0,2
0,0
0,0
-0,2
-0,2 -0,4
14 14 14 14 14 14
0 1x10 2x10 3x10 4x10 5x10 6x10 0 1x10
14 14
2x10 3x10
14 14
4x10 5x10
14 14
6x10
dose [ions / cm²] dose [ions / cm²]
Fig. 2: Dependence of the bias field and the coercivity from the Ga+ -ion dose. The data of the bias field was
fitted with a model function, which is plotted in both diagrams.
AG Magnetismus
74
TU Kaiserslautern
Fig. 3: Energy losses of the ions as a function of the penetration depth in units of Ångstrøm. The left diagram
shows the energy losses for the 5 keV He+ -ions, while the right diagram shows them for the 30 keV Ga+ -ions.
It is obtained that the coercivity exhibits a behavior which is very similar to that of the bias field.
Moreover, one can conclude from Fig. 2 that the enhancement of the bias field in this system for
the irradiation with 5 keV He+ -ions is an ion-specific effect.
To investigate these phenomena further, TRIM-simulations have been performed, using a numer-
ical simulation program written by Ziegler, Biersack and Littmark [3]. The irradiation of the
investigated layer system with 5 keV He+ -ions and 30 keV Ga+ -ions was simulated. Figures 3 and
4 show some of the results of this simulation.
A notable difference between the two irradiation procedures is observed. In the case of He+ -ions
the main channel for energy loss is the ionization of the target atoms, while in the case of Ga+ -ions
most of the ion energy is lost due to dislocation of the target atoms. A plot of the defects created
in the target material shows a maximum in the region of the ferromagnet-antiferromagnet interface
in case of gallium, while for helium the creation of defects is more uniform. A special plot of
this interface shows that the intermixing, i.e. the dislocation of Ni atoms into the antiferromagnet
Fig. 4: Defects created during irradiation per ion as a function of the penetration depth in units of Ångstrøm.
The left diagram shows the defects created by 5 keV He+ -ions, while the right diagram shows them for 30 keV
Ga+ -ions.
AG Magnetismus
75
TU Kaiserslautern
and the dislocation of Mn atoms into the ferromagnet, is considerably higher for the same dose of
gallium than for helium, approximately by a factor of 40. This is consistent with the result that a
comparable decrease of the bias field for the irradiation with Ga+ -ions occurs at doses which are
approximately 60 times lower than in the case of He+ -ion irradiation.
In summary, it is obtained that the irradiation with Ga+ -ions is much more efficient in the creation
of defects than the irradiation with He+ -ions, and in the case of gallium those defects are primarily
created in the interface region between ferromagnet and antiferromagnet. These results support
the model outlined above, which attributes a reduction of the bias field to the creation of defects
within this interface region.
References
[1] S. Poppe, Modifikation des Austausch-Verschiebungseffektes in Ferro-/Antiferromagnet-Bilagen durch lagen-
weise und ortsaufgelöste Ionenbestrahlung, Dissertation, TU Kaiserslautern (2003).
[2] J. Juraszek, J. Fassbender, S. Poppe, T. Mewes, B. Hillebrands, D. Engel, A. Kronenberger, A. Ehresmann, H.
Schmoranzer J. Appl. Phys. 91, 6896 (2002).
[3] J.F. Ziegler, J.F. Biersack, J.P. Littmark, The Stopping and Range of Ions in Solids: Pergamon, New York
(1985).
AG Magnetismus
76
TU Kaiserslautern
E. Elastic Properties
6.14 Growth dynamics in sputtered BN films revealed by Brillouin light
scattering
T. Wittkowski, K. Jung, and B. Hillebrands1
Boron nitride (BN) in its cubic crystallographic modification is superior to diamond in many as-
pects concerning electronic and wear protective applications. The understanding of the mecha-
nisms, which lead to the growth of cubic BN thin films, is a worldwide subject of investigation [1].
For the deposition on almost all substrate materials the initially developing sp2 -bonded BN layer
at the substrate interface, which may possess a thickness of a few tens of nanometers, plays a key
role for the nucleation of the cubic phase and for the elastic properties of the layered film system
as a whole. The aim of this project is to reveal the microstructure of specially prepared hexagonal
boron nitride (h-BN) films by Brillouin light scattering (BLS) and HRTEM as analytical methods.
Films were deposited by r.f. magnetron sputtering of an h-BN target after sputter cleaning the
Si(001) substrate. The substrate temperature was 350 ◦ C. The 4” magnetron was operated at
13.56 MHz with an r.f. power of 800 W. The working gas pressure was 0.2 Pa with a gas flow ratio
of Ar/N2 = 97/3, leading to stoichiometric BN films [2]. The electric potential of the substrate
was kept at 0 V during the deposition like the chamber walls. Films of different thicknesses were
produced by variation of the deposition time in several deposition runs under identical conditions.
BLS investigations were carried out in 180 ◦ -backscattering geometry in (p − ps)-polarization
at room temperature [3]. The light power at the sample surface was 120 mW; a collecting lens
of numeric aperture 1.7 was used. For surface excitations the relations for the parallel phonon
wavevector and the energy are as in Sect. 6.15.
The propagation of acoustic waves in thin films is described within the framework of elastic contin-
uum theory. For a single thin film on a substrate, acoustic modes are guided in the film material if
Fig. 1: Dispersion of the phase

7000 h = 45 nm velocity of vertically polarized
h = 270 nm
h = 500 nm surface phonons, guided in the h-
6000 BN films. The diagram shows
phase velocity [m/s]
the combined experimental data

5000 of three films of different thick-
ness h. The parallel phonon
4000 wavevector q was varied by the
angle of light incidence for each
film. Pronounced discontinuities
3000
in the sound velocities of films of
different thickness and identical
2000 mode order are observed. This
effect is apparent for q · h ≈ 6.
0 2 4 6 8 10 12 The dashed line indicates the cut-
q||h off velocity of the silicon sub-
strate.
1 In collaboration with T. Pfeifer, I. Hermann, and F. Richter, Institut für Physik, TU Chemnitz, Chemnitz, Ger-
many, and T. Chudoba, ASMEC Advanced Surface Mechanics, Dresden, Germany.
AG Magnetismus
77
TU Kaiserslautern
their sound velocities are smaller than that of the substrate, as is the case for the investigated h-BN
films. If the film thickness is in the range of the surface phonon wavelength or smaller, the phase
velocity of the guided modes is dispersed. The velocities of the dispersion curves depend on the
material parameters and on the dimensionless product of the surface phonon wavevector q and the
film thickness h [4]. Film elastic properties are described by so-called effective elastic constants
since in the measurement microscopic film properties, applying to each of the h-BN nanocrys-
tallites, are averaged over a sufficiently large lateral extension of at least the phonon wavelength
(varied from 0.27 to 0.8 µm). In order to describe film texture and possible growth structures by
the effective elastic constants, it is necessary to assume a hexagonal symmetry of the film with
the c-axis parallel to the film normal. Solutions of the equation of motion with vertical and lon-
gitudinal displacement components (Rayleigh mode and Sezawa modes) decouple from solutions
with shear-horizontal components for such a film on a (001) substrate. In consequence the descrip-
tion of the vertically polarized modes requires only four independent components of the stiffness
tensor, which are c11 , c13 , c33 , c55 , and the mass density [4].
Figure 1 shows the experimentally determined dispersion of three h-BN films of different thick-
ness. As was pointed out above, the dispersion of surface acoustic modes should be plotted as
continuous curves, these being functions of q · h. However, if the film material alters its mi-
crostructure as a function of the distance from the interface, the dispersion depends on q and h
independently. This results in discontinuities in the velocity dispersion of films of different thick-
ness for identical values of q · h. Such a behavior is evidenced by the experimental data shown
in Fig. 1. In consequence it must be concluded that the investigated h-BN films change their mi-
crostructure with growing film thickness. These findings are corroborated by a HRTEM analysis
from the film cross section. As can be seen in Fig. 2 the nanocrystalline material alters its crystal-
lite orientation distribution function in dependence of the distance from the interface drastically.
The film grows with a strict fibre texture up to thickness of a few 10 nm. In the further growth
the texture becomes weaker until it is completely lost. Films of thickness 200 nm or more are
essentially not textured, i.e. their crystallite orientation distribution function is a constant.
For this type of films, characterized by a texture gradient, it is impossible to specify a single set
of elastic constants which describes properly the stiffness for all samples independent of their
thickness [5]. Nevertheless the rich experimental data base can be evaluated quantitatively. For
this purpose the measured dispersion is fitted by solving the equation of motion numerically with
the stiffness constants of the film as free parameters. Figure 3 shows an enlarged clipping of Fig. 1
containing experimental data and calculated velocity dispersion curves which represent the least
squares fitting result. Error bars of the individual measurements are taken into account by using
weights in the computation of the χ2 -sum. The selected experimental velocities, which were used
for the fit, correspond to the short-wavelength surface phonons and to the Rayleigh mode of the
500 nm thick film. Thus the obtained elastic constants describe the film stiffness at the surface
rather than at the interface. The fitting result is c11 = 37.1 ± 3.2 GPa, c33 = 59.7 ± 3.4 GPa, and
c55 = 6.5 ± 0.2 GPa with c13 = 24.0 GPa and ρ = 1.61 g/cm3 . Since c13 is correlated with the other
parameters, it was calculated from the other constants in each iterating step of the fit procedure
using the relation 2c55 /(c11 − c13 ) = 1 (i.e. elastic isotropy). The χ2 -sum is little affected by
the value of c13 . Fixing the constant c55 at the above value and fitting the mass density yields
ρ = 1.61 ± 0.05 g/cm3 in agreement with the result of an independent determination by X-ray
reflectometry. The χ2 -sum divided by the number of degrees of freedom is 1.60 thus documenting a
fair reliability of the fit. However, systematic errors arising from uncertainties in the film thickness
determination and from the finite aperture size of the collecting lens limit the accuracy of the
deduced constants to ±5 %.
AG Magnetismus
78
TU Kaiserslautern
Fig. 2: HRTEM micrographs of the cross section

of a thick h-BN film. The lower graph shows the
microstructure at a distance of 50 nm from the in-
terface. The pronounced fibre texture is charac-
terized by the crystallite c-axes lying in the film
plane. The upper graph shows the microstructure
at a distance of 400 nm from the interface. The h-
BN crystallites are a few nanometers in diameter
and are randomly oriented. The arrows indicate
the direction of the film normal.
In order to investigate the variation of film stiffness as a function of the distance from the interface
quantitatively, the combined experimental data shown in Fig. 1 of all three films are fitted in order
to obtain a set of elastic constants which describe the stiffness of these films as a whole. Data points
for which the assignment to the correct mode order was ambiguous were omitted. The remaining
data set consisting of 74 data points is still particularly large. It must be noted, however, that the
reliability of this fit result is rather poor due to the failure of the model which treats the films as
an effective medium for acoustic wave propagation, and thus assumes that the film properties, as a
function of the distance from the interface, are constant. Transferring the constant c13 = 24.0 GPa
and ρ = 1.61 g/cm3 from the afore obtained fit result, a reduction of c11 and c55 and an increase of
c33 is obtained from the combined data of all samples. These results provide an indication of how
the effective stiffness of this type of h-BN films changes with thickness.
Apparently the elastic anisotropy, which is expressed by the anisotropy factor A = c11 /c33 , is
smaller than 1. The fit result of the 500 nm thick films yields A = 0.62, whereas for all three
films combined the anisotropy is even more pronounced. In earlier studies it was shown that
the anisotropy in this type of film is caused by the texture and has its origin in the enormous
anisotropy of the h-BN single crystal [6]. The anisotropy weakening of the thickest sample is in
full agreement with the thickness dependent variation of the microstructure as revealed by TEM
in Fig. 2. It is concluded that the thicker these films are, the more justified is their treatment as
elastically isotropic.
The Young’s modulus E parallel to the film normal is calculated from the elastic constants. We
set c12 = c13 and regard that c11 > |c12 |. From the fit of the thickest sample it follows that E =
41 ± 5 GPa. Using the combined data of all samples one obtains an even higher modulus.
AG Magnetismus
79
TU Kaiserslautern
Fig. 3: Dispersion of the

Rayleigh mode and the Sezawa
5000 modes of a 500 nm thick h-BN
film in the discrete part of the
mode spectrum. Experimental

4000 data (triangles) corresponding to
surface phonons with wavevec-
tors q > 1.95 · 105 cm−1 and
3000 of the Rayleigh mode are used
for a least squares fit of the film
elastic constants. The lines show
2000 the best fit result.
2 4 6 8 10 12
q||h
In summary this study points out the high sensitivity of the phase velocity of guided acoustic modes
on the microstructure of layered structures, measured by BLS. As was shown for h-BN films,
small alterations, mainly in the film texture, clearly show up as discontinuities in the combined
velocity dispersion curves of differently thick films. Thus BLS from surface phonons in the GHz
frequency range allows for a quantitative insight into the effective elastic behavior of thin films
as a function of the distance from the interface. These results are in agreement with those of an
independent technique using a spherical indenter [7]. The combination with the findings of several
other analytical techniques such as TEM and X-ray reflectivity yields a self-consistent description
of elastic and microstructural film properties.
References
[1] X.W. Zhang, H.G. Boyen, N. Deyneka, P. Ziemann, F. Banhart, M. Schreck, Nature Mat. Lett. 2, 312 (2003).
[2] J. Hahn, F. Richter, R. Pintaske, M. Röder, E. Schneider, T. Welzel, Surf. Coat. Technol. 92, 129 (1997).
[3] B. Hillebrands, Rev. Sci. Instrum. 70, 1589 (1999).
[4] G.W. Farnell, E.L. Adler, in Physical Acoustics, Vol. 9, edited by W.P. Mason and R.N. Thurston (Academic,
New York, 1972), pp. 35-127.
[5] T. Wittkowski, V. Wiehn, J. Jorzick, K. Jung, B. Hillebrands, Thin Solid Films 368, 216 (2000).
[6] T. Wittkowski, P. Cortina, J. Jorzick, K. Jung, B. Hillebrands, Diam. Rel. Mat. 9, 1957 (2000).
[7] T. Chudoba, N. Schwarzer, F. Richter, Surf. Coat. Technol. 127, 9 (2000).
AG Magnetismus
80
TU Kaiserslautern
6.15 Evidence of surface phonons in mesoscopic BN coated fibres
T. Wittkowski, K. Jung, and B. Hillebrands1
The reinforcement effects of fibres in metal or ceramic matrix composites are widely influenced
by suitable coatings. The aim is to facilitate an optimized fibre/matrix adhesion as well as sliding
of fibres in the matrix under mechanical load. Moreover such coatings might serve as a diffusion
barrier protecting fibres against oxidation and other corrosion processes. In this report, a boron ni-
tride coating is used as a promising alternative to pyrolytic carbon due to its graphite-like structure,
allowing for good fibre/matrix sliding properties and excellent oxidation resistance.
The investigated fibres are of 5 microns in diameter and thus in a mesoscopic region of size. They
are of dimensions which are between those of other objects of cylindrical symmetry studied so
far by light scattering [1, 2]. The observation of surface acoustic excitations provides insight into
their elastic properties, and, in particular, allows one to measure the thickness, the mass density
and the elastic properties of thin film coatings deposited on bundles of fibres. The non-destructive
measurement of such properties seemed to be impossible up to now. Here we report the successful
elastic characterization of such structures.
Carbon fibres T 800 from “TORAYCA” were employed consisting of 6000 filaments - each 5 µm
in diameter, density 1.81 g/cm3 , tensile modulus 295 GPa. The fibres have been coated with BN in
a continuous process via an isothermal chemical vapor deposition process in a vertical hot wall re-
actor at 1100 ◦ C under atmospheric pressure [3]. The composition of the BN films deposited onto
silicon wafers under analogous conditions was measured by wavelength dispersive X-ray spec-
troscopy (EPMA, CAMECA SX 100). The N/B ratio was found almost stoichiometric or some-
what greater than 1. The density of the deposits was measured in a mixture of 1,2-dibromethan and
aceton using the sink-float method for both coating pieces cracked off the reactor wall and coated
fibres compared with uncoated ones. The values spread in the range of ρ f ilm = 1.88 ± 0.02 g/cm3 .
In comparison to the specifications provided by the manufacturer the density of the uncoated fibres
has also been measured yielding ρ f ibre = 1.809 ± 0.001 g/cm3 . Transmission electron microscopy
studies (HREM diffraction) were performed using a 200 kV HITACHI H-8100. HREM images
of cross sections reveal the hexagonal turbostratic structure of the BN layer. The observed high
structural perfection is limited to small volumes forming nanocrystals or cellular structures of only
Fig. 1: Sample coordinate system and scheme

ks of the scattering geometry. ki and ks are the
ki wavevectors of the incident and the scattered
light, q is the parallel phonon wavevector in a
q|| Stokes process. The scattering plane is defined
x1 by the x̂1 and the x̂3 -axis in the Cartesian sample
coordinate system. The detected parallel phonon
R
wavevector is selected by the angle of incidence
θ. The fibre radius R is up to 10 times larger than
x3
the wavelength of the acoustic excitations.
x2
1 Incollaboration with S. Stöckel, K. Weise, and G. Marx, Institut für Chemie, Physikalische Chemie, TU Chem-
nitz, Straße der Nationen 62, D-09107 Chemnitz.
AG Magnetismus
81
TU Kaiserslautern
some nanometers in diameter. These nanocrystals consisting of aligned (002) BN hexring sheets
are commonly oriented more or less randomly, apart from the very first lattice planes at the carbon
interface where the hexagonal planes are oriented parallel to the fibre surface. The TEM analysis
also yielded the film thickness of some of the outermost fibres of the bundle. By additional de-
termination of the mass difference of the coated and the uncoated fibres the film thickness on the
outer bundle fibres was found to be h = 80 ± 15 nm.
BLS experiments were performed with the fibre axes lying in the sagittal plane. A special sample
holder was developed which guarantees that all fibres of a bundle were fixed parallel to each other.
The achieved fibre alignment was accurate within a margin of ±5 ◦ for the uncoated carbon fibres
and somewhat lower for the BN coated fibres. Figure 1 provides a schematic diagram of the
scattering geometry. The scattering plane contains the x̂1 - and the x̂3 -axis in the sample coordinate
system. The fibre length is infinite in comparison with the mean free phonon path which is in the
order of a few microns. The fibre radius R is 2.5 µm so that R q > 2π; q denotes the phonon
wavevector parallel to the fibre axis.
1. Results on carbon fibres

A strictly linear behavior of the phonon frequency as a function of sin θ is observed which proves
that the phonon wavevectors involved in the scattering process are parallel to the x̂3 -axis. Since
no elastic boundary conditions exist for the x̂3 -direction, wavevector conservation is valid for the
parallel component yielding
q = 2ki sin θ . (1)
With the energy conservation law of the Brillouin process the angular frequency of the associated
acoustic excitation of phase velocity v writes
Ω = v · q . (2)
In a least squares fit the measured phase velocity is determined to v = 2840 ± 10 m/s from the
slope of the data points.
Since the fibre diameter is several times larger than the acoustic wavelength, the fibre geometry
does not impose strict boundary conditions in the continuum mechanical description of phonon
states. In addition the opacity of the graphitic material confines the interaction of light with
phonons to a near-surface region of the fibres. Consequently the observed excitation is identified
with the Rayleigh surface mode which is detected due to the corrugation of the fibre surface. Up
to now the Rayleigh mode of graphite has been observed only for planar sample geometries [5–7].
In order to relate the measured phonon frequencies with the elastic properties, each fibre is treated
as an effective medium possessing hexagonal symmetry concerning its acoustical properties; the
x̂3 -fibre axis is oriented parallel to the hexagonal c-axis. This provides for a different elastic behav-
ior parallel and perpendicular to the fibre axis. Due to the production process it is expected that the
stiffness in direction of the axis is much higher than in perpendicular directions. In order to obtain
such properties, which also lead to a high tensile strength, fibres were strained and annealed in the
final production steps. The hexagonal symmetry describes the elastic properties to be identical in
all directions perpendicular to the fibre axis. It is noted that the measured effective hexagonal sym-
metry is not directly related to the orientation distribution function of graphitic crystallites possibly
present in the fibres.
Results are discussed in the limit of a planar geometry where R q 2π, so that the fibre material
forms an infinite halfspace for the Rayleigh mode. For the sample frame of reference (see Fig. 1)
AG Magnetismus
82
TU Kaiserslautern
Fig. 2: Brillouin spectrum

of carbon fibres coated
Rayleigh
16 with 80 nm h-BN. The an-
scattered intensity [a.u.]
gle of incidence is θ =
58.2 ◦ with the scattering
plane containing the fibre
12 axis. A broad resonance
in the continuous part of
the spectrum shows up at
±10.5 GHz. The Rayleigh
resonance
mode is the dominant fea-
8 ture in all measured spec-
tra. The central peak,
which is due to elastically
scattered light, has been re-
4 moved for clarity.
-12 -8 -4 4 8 12
frequency shift [GHz]
the surface solution of the equation of motion under the elastic boundary conditions of the free
surface is obtained from the solution propagating on the basal plane. The stiffness tensor, however,
has to be rotated in order to account for the transformation from the sample into the laboratory
frame of reference. The Rayleigh velocity vR is obtained from the real positive root of an implicit
cubic equation in v2R . It depends on the mass density as well as on the elastic constants c11 , c13 ,
c33 , and c44 = c55 of the fibre material [4]. As is well known vR is predominantly determined by
the shear modulus c55 . This fact has been used in the past, e.g. to deduce the shear modulus of
s.c. graphite [5, 6]. An exemplary computation using the known elastic constants of graphite [7, 8]
shows that for propagation
on a surface in direction
of the x̂3 -axis c55 /cR − 1 = 7.7 · 10−4 , with the
Rayleigh velocity vR = cR /ρ and vT1,2 = c55 /ρ. The phase velocities vT1,2 of the two transverse
bulk branches degenerate for this particular direction, ρ denotes the mass density.
f ibre
With vR = 2840 m/s the Rayleigh velocity of the fibres is almost a factor of two higher than
graphite
for a s.c. graphite material (vR = 1494 m/s) with the same symmetry properties. Using the
f ibre
mass density of the fibres their shear modulus amounts to c55 14.60 ± 0.11 GPa. For compar-
s.c.graphite
ison, the shear modulus of s.c. graphite is c55 5.05 ± 0.35 GPa [7] and for high-ordered
polycrystalline graphite (HOPG), c55 HOPG 3.25 ± 0.15 GPa [6]. The above results reveal a fibre
shear stiffness which is between the ones of graphite and those of partially hydrogen containing,
amorphous sp2 -bonded carbon films [9, 10]. It is remarkably lower than the shear moduli found
for amorphous tetrahedrally bonded films (ta-C) [11].
2. Results on BN coated carbon fibres
In general the mass density and the stiffness tensor of a film material deposited onto a substrate
differs from that of the substrate. As a consequence the velocities of surface acoustic modes show
a characteristic dispersion, as long as the film thickness is smaller than the wavelength of the
excitations. This is exactly what is observed in the BN coated fibres (Fig. 3). The phase velocity
of the Rayleigh mode decreases distinctly with increasing value of q h, the product of the parallel
phonon wavevector and film thickness h. In the limit of q h → 0 the phase velocity is that of the
f ibre
Rayleigh velocity of the uncoated fibre, vR , whereas in the limit of q h → ∞ it is the Rayleigh
AG Magnetismus
83
TU Kaiserslautern
4000 Fig. 3: Phase velocity

of surface excitations in
BN coated carbon fibres
3500 vs. the product of paral-
lel phonon wavevector q

and BN film thickness h
3000
(here h = 80 nm).
fibre
vR = 2840 m/s The
sound velocity of the film is
lower than that of the sub-
2500 f ibre
strate, vR = 2840 m/s
(solid line). From the dis-
2000 persion curve of the low-
est mode, the Rayleigh
f ilm
vR
film
≤ 1820 m/s mode, it is seen that vR ≤
1500 1820 m/s can be used as an
0.0 0.4 0.8 1.2 1.6 2.0 upper limit for the Rayleigh
velocity of the film (dashed
q||h (h = 80 nm)
line).
f ilm
velocity of the film material, vR . The presence of an additional higher-order Rayleigh-like mode,
a so-called Sezawa mode, is weakly indicated in two spectra. Surface resonances are observed in
the continuum of excitations in a few of the experimental spectra, one of which is shown in Fig. 2.
The corresponding sound velocities are also shown in the dispersion in Fig. 3. For a few values of
q the measurement was repeated at different positions on the fibre bundle which gives rise to some
scatter in the velocity data. This scatter is partially attributed to an uncertainty in the homogeneity
of the fibre orientation during the measurement. It is however predominantly caused by a film
thickness which varies for the different fibres in the bundle.
From Fig. 3 it is clear that the lowest measured velocities should make up an upper limit for the
f ilm
Rayleigh velocity of the BN film material. Thus vR ≤ 1820 m/s and, following the discussion for
f ilm
the carbon fibres, c55 ≤ 6.23 GPa with the mass density of the film, ρ f ilm = 1.88 g/cm3 . Such a
value of the BN shear modulus is in agreement with its s.c. elastic constants which are comparable
with those of graphite [12–14]. These findings prove a reduction of the shear modulus in the near-
surface region of the coated in comparison to the uncoated fibre by more than a factor of two.
This is likely to improve the fibre properties significantly if they are in mechanical contact with
each other and with the matrix material in fibre-reinforced composites. The damaging influence
of shear stresses at the fibre-matrix interfaces should be reduced by the lubricating effect of the
high-temperature resistant BN coating without any reduction of the fibre tensile strength. It is
expected that the application of such coated fibre bundles leads to an increased fracture toughness
of technically highly relevant composite materials at elevated temperatures.
This is the first time that the BLS technique is applied to analyze elastic properties of fibres which
possess a few microns in diameter. These properties are difficult or even impossible to determine
with other techniques. BLS does not modify material properties during the measurement and it is
absolutely non-destructive. The alteration of near-surface mechanical properties due to the BN film
is clearly evidenced and shows the potential of BLS for the characterization of elastic constants,
mass density, and the film thickness. A full report is given in [15].
AG Magnetismus
84
TU Kaiserslautern
References
[1] J.W. Palko, A. Sayir, S.V. Sinogeikin, W.M. Kriven, and J.D. Bass, J. Am. Ceram. Soc. 85, 2005-12 (2002).
[2] C.E. Bottani, A. Li Bassi, M.G. Beghi, A. Podesta, P. Milani, A. Zakhidov, R. Baughman, D.A. Walters, and R.
E. Smalley, Phys. Rev. B 67, 155407 (2003).
[3] K. Weise, S. Stöckel, and G. Marx, in Werkstoffwoche; Vol. 98, edited by W. Dimigen and W. Paatsch (Wiley-
VCH, Weinheim, 1999), p. 121-6.
[4] M. J. P. Musgrave, Crystal Acoustics (Holden-Day, San Francisco, Cambridge, London, Amsterdam, 1970).
[5] M. Grimsditch, J. Phys. C (Solid State Physics) 16, L143-4 (1983).
[6] S.A. Lee and S.M. Lindsay, Phys. Status Solidi (b) 157, K83 (1990).
[7] M. Grimsditch, Phys. Status Solidi (b) 193, K9 (1996).
[8] O.L. Blakslee, D.G. Proctor, E.J. Seldin, G.B. Spence, and T. Weng, J. Appl. Phys. 41, 3373-82 (1970).
[9] T. Wittkowski, V. Wiehn, J. Jorzick, K. Jung, and B. Hillebrands, Thin Solid Films 368, 216-21 (2000).
[10] X. Jiang, Phys. Rev. B 43, 2372-7 (1991).
[11] A.C. Ferrari, J. Robertson, M.G. Beghi, C.E. Bottani, R. Ferulano, and R. Pastorelli, Appl. Phys. Lett. 75,
1893-1895 (1999).
[12] J.F. Green, T.K. Bolland, and J.W. Bolland, J. Chem. Phys. 64, 656 (1976).
[13] L. Duclaux, B. Nysten, J.-P. Issi, and A.W. Moore, Phys. Rev. B 46, 3362 (1992).
[14] R.W. Lynch and H.G. Drickamer, J. Chem. Phys. 44, 181 (1966).
[15] T. Wittkowski, K. Jung, B. Hillebrands, S. Stöckel, K. Weise, and G. Marx, submitted to Anal. Bioanal. Chem.
(2003).
AG Magnetismus
85
TU Kaiserslautern
6.16 Monte Carlo method to analyze the dispersion of guided acoustic modes
T. Wittkowski, G. Distler, K. Jung, and B. Hillebrands
Surface acoustic waves are of high value for the characterization of elastic and microstructural
properties of thin films. The evaluation of the dispersion of guided modes in thin films in the GHz
frequency range, as accessible with Brillouin light scattering (BLS), allows for a detailed descrip-
tion of anisotropic film properties. Usually material constants are extracted from a fit procedure in
which the elastic constants act as free parameters [1]. For this the sum of squares
χ2 = ∑ wi (vcalculated
i − vmeasured
i )2 (1)
i
over i measured and calculated velocities, using the weights wi , is mostly used as the merit func-
tion. It is minimized in consequent iteration steps by a deterministic procedure. This technique
is well established, it works fast and reliably provided that the assignment to the correct acoustic
mode order is clear. For convergence, these algorithms, the Levenberg-Marquardt method being
a prominent example, usually need start values for the fit parameters not too far away from the
solution corresponding to the minimum sum of squares, χ2min , without guarantee that the absolute
minimum is found. The structure of the parameter space bears additional information about pa-
rameter correlation and uniqueness of the found solution. Commonly used deterministic fitting
routines assume a normal distribution of the parameters. If this is not fulfilled − a situation which
appears quite regularly − confidence limits on parameters computed by using such routines provide
only an approximation. We thus adopted an alternative statistical Monte Carlo (MC) technique in
order to address some of the above named issues.
The MC procedure shall be described briefly. Starting from the experimental data, new data sets
consisting of the same number of data points were generated by a random variation of quintuplets
of the free parameters, four independent constants of ci j and ρ, within specified intervals. The
number of parameters may be increased or reduced depending on the symmetry properties of the
system to be described. Here we consider sagittal modes in a hexagonal symmetry film. The
weighted χ2 -sum for each data set was computed according to Eq. (1). The velocities vcalculated
i
were obtained from the zeros of the boundary condition determinant which contains the compo-
nents of the elastic tensors and the mass densities of both the film and the substrate, coupled by the
Fig. 1: Phase velocity dispersion

5500 curves for the discrete, sagit-
tally polarized surface waves
5000 (Rayleigh and Sezawa) as a

4500 function of q · h. The filled
circles represent measurements
4000 made with a constant q value
of 17.27 µm−1 over the complete
3500 range of film thickness. For two
film steps of thickness 191 nm
3000
(open squares) and 595 nm (open
2500 triangles) the value of q was var-
ied. The lines represent the com-
0 2 4 6 8 10 12 puted best fit to the experimental
q||h data.
AG Magnetismus
86
TU Kaiserslautern
460
(a)
Fig. 2: Projections of the parameter space onto the
(c11 , c33 )-, the (c33 , c13 )-, and the (c55 , ρ)-plane
440
420
for the quintuplets of free parameters with χ2 in
400
the vicinity of χ2min . The three chosen planes rep-
c33 [GPa]
380
360
resent the parameter pairs which are most strongly
340
2
∆χ ≤ 6 correlated. Projections of different sets of points
320
2
∆χ ≤ 4 show areas that are related to different values
∆χ ≤ 2
of ∆χ2 , where ∆χ2 = χ2 − χ2min . The constant
2
300
320 330 340 350 360 370 380 390 400 c13 , which usually is difficult to determine due to
c11 [GPa] its strong correlation with the other constants, is
200 quantified with fair accuracy.
(b)
180
160
c13 [GPa]
140
120
100
300 320 340 360 380 400 420 440 460

c 33 [GPa]
13.8 (c)
13.6
13.4
13.2
ρ [g/cm ]
3
13.0
12.8
12.6
12.4
12.2
106 108 110 112 114 116 118 120 122
c55 [GPa]
boundary conditions at the interface

f ilm f ilm
σsubstrate
3j = σ 3j and usubstrate
3 = u3 (2-a)
x3 =0 x3 =0
and at the free surface [3]

f ilm
σ3 j =0 . (2-b)
x3 =h
The actual set of parameters is retained if the calculated dispersion curves fall − for 68 % of
all points of the set in this example − within the error bars of the measured data points. Other
parameter sets are discarded. This routine does not require any assignment to the mode order.
The analysis was performed on the dispersion of surface acoustic modes in tungsten carbide films
on silicon used as a model system [2]. In addition the large velocity dispersion data set allowed
the determination of four independent stiffness constants as well as the mass density of the film
material by a fitting routine. The measured dispersion is shown in combination with the result of
the least squares fit in Fig. 1.
Since the deterministic fitting algorithm converged satisfactorily in determining the minimum in
the χ2 -sum, we concentrated our statistical analysis on the parameter space surrounding of the
absolute minimum χ2min . Figures 2 and 3 present the results of 7.5 · 108 randomly chosen and
evaluated data sets. Computation time on a modern personal computer with a processing frequency
AG Magnetismus
87
TU Kaiserslautern
30
Fig. 3: MC results of the weighted sum of squares
for the two anisotropy factors A1 and A2 of the
28
film material. Starting from the hexagonal sym-
sum of squares
26 metry model, the smallest χ2 -sum clearly corre-

24 sponds to nearly isotropic parameters for this type
of film. The straight line in both diagrams indi-
cates the minimum χ2 -sum that was obtained with
22
χ
2
20 min the least squares fit.
0.85 0.90 0.95 1.00 1.05 1.10
anisotropy A1 = c11/c33
30
28
sum of squares
26
24
22
χ
2
20 min
0.95 1.00 1.05 1.10 1.15 1.20

anisotropy A2 = 2c 55 / (c 11-c13)
of 1.8 GHz was approximately one week. The computation can be distributed on several CPUs to
shorten this time.
Figure 2a-c shows projections of the parameter space onto the (c11 , c33 )-, the (c33 , c13 )-, and the
(c55 , ρ)-plane for the free parameter quintuplets with χ2 in the vicinity of χ2min . Among the 10
two-dimensional subspaces these three chosen planes represent the parameter pairs which are the
most strongly correlated ones. Projections of different sets of points show areas that are related
to different values of ∆χ2 , where ∆χ2 = χ2 − χ2min . These diagrams emphasize the parameter
correlation and provide an insight into the structure of the parameter space. A detailed analysis of
the probability distribution functions of the individual parameters shows that these are not normally
distributed in general so that the results shown in Fig. 2 provide an appropriate presentation of the
parameter space for chosen values of ∆χ2 . Figure 3 shows the MC results of the weighted sum of
squares for the two anisotropy factors A1 and A2 of the film material. Starting from the hexagonal
symmetry model, the smallest χ2 -sum clearly corresponds to nearly isotropic parameters.
In summary the least squares fitting and the MC approach yield self-consistent results on four
independent components of the stiffness tensor and the mass density of the tungsten carbide film.
Both techniques reveal a near isotropy of the film material by using a hexagonal model. The
statistical MC method is well appropriate to determine the absolute and relative minima of a merit
function yielding additional information about the often complex structure of a multi-dimensional
parameter space. A full report is given in [2].
References
[1] W.H. Press, S.A. Teukolsky, V.T. Vetterling, B.P. Flannery, Numerical recipes, 2nd ed., Cambridge University
Press (1992).
[2] T. Wittkowski, G. Distler, K. Jung, B. Hillebrands, J.D. Comins, submitted to Phys. Rev. B (2003).
[3] G.W. Farnell, E.L. Adler, in Physical Acoustics, Vol. 9, edited by W.P. Mason and R.N. Thurston (Academic,
New York, 1972), pp. 35-127.
AG Magnetismus
88
TU Kaiserslautern
F. Transfer of Technology
6.17 Mode-locked ps-Nd:YVO4 - laser for time-resolved measurements at
wavelengths of 1064 nm, 532 nm and 266 nm
M.C. Weber, H. Nembach, J. Fassbender1
In this report we describe our new picosecond laser setup which was developed in collaboration
with Lumera Lasers GmbH [1].
Ultra-short laser pulses are a powerful tool for time resolved investigations of magnetic films.
The benefits of this method have been demonstrated so far mostly with Ti:Sapphire lasers, which
provide pulse durations shorter than hundred femtoseconds and enable the observation of ultrafast
processes such as carrier interactions. However, such Ti:sapphire lasers are complex and expensive
devices, which are limited in pulse energy. In contrast, picosecond lasers, based on Nd:YVO4 , offer
the advantage of direct diode pumping and provide high efficiency and high average power from a
robust system. The pulse duration of ≈ 10 ps provides the right time scale for many experiments,
in which the minimum time scale is defined by the duration of magnetic field pulses generated by
an electronic pulse generator of ≈ 100 ps used to excite the magnetic system.
The Nd:YVO4 laser described here features high pulse energy in the infrared and a subsequent
frequency conversion to green and ultraviolet (UV) radiation. Single pulses or pulse trains can
be selected with a pulse repetition frequency from single pulses up to 1 MHz. The variety of
wavelengths and repetition rates empowers the user to drive several experiments with one laser
source. A special feature is the optical generation of trigger pulses with minimum timing jitter.
A scheme of the laser setup is shown in Fig. 1. Pulses are generated at a wavelength of 1064 nm
by a passive mode-locked laser oscillator with a duration of 8.5 ps and a pulse repetition frequency
(PRF) of 80 MHz. The pulse energy is Elaser > 50 nJ, the average output power is Plaser = 4.3 W.
A double-pass amplifier (AMP) maintains a perfect beam quality (M 2 < 1.1) and increases the
output power to PIR = 7.5 W (EIR = 90 nJ). These pulses are focused into a nonlinear LBO crystal2
PS-LASER AMP SHG AOM2 FHG
AOM1 / TRG
port 2 port 3 port 4

wavelength: 1064 nm 532nm 532nm / 266nm
PRF: 80 MHz 80 Mhz typ. 100 kHz
pulse energy 25 nJ 50 nJ 35 nJ / 3 nJ
Fig. 1: Scheme of the Nd:YVO4 laser setup for generation of picosecond pulses in the IR, green, and UV. For
the nomenclature see the main text.
1 in collaboration with R. Knappe, T. Herrmann, B. Henrich, A. Nebel, Lumera Lasers GmbH.

2 SHG: Lithiumtriborat: LiB3 O5 (LBO) T = 150 ◦ C, cut φ = 0 ◦ , θ = 90 ◦ .
AG Magnetismus
89
TU Kaiserslautern
a) b)
1.0 Pout = 4.3 W Sech2-Fit

M2 < 1.1
0.8
0.6
rel. Intensity
2
Sech
=8.5 ps
0.4
0.2
0.0
-30 -20 -10 0 10 20 30
Delay [ ps ]
Fig. 2: a) Transmission window (upper line) for a selected laser pulse (lower line), b) Intensity autocorrelation
of the IR laser pulses.
to generate the second harmonic (SHG) with a conversion efficiency > 50 percent (Pgreen = 4 W,
Egreen = 50 nJ). The residual IR is available with an output power of 2 W (25 nJ). A small portion
of the green beam (≈ 10 mW) is split to generate the trigger signal (TRG). In both laser beams,
acousto-optic modulators (AOM) are used as pulse pickers with a tunable pulse repetition fre-
quency. The efficiency of the AOM is ≈ 70 percent and limits the energy of the selected green
pulses to 35 nJ. These pulses are frequency-doubled again in a nonlinear crystal3 and generate the
fourth harmonic (FHG) with a wavelength of 266 nm and a pulse energy of > 3 nJ.
A critical issue for time-resolved measurements is a low-jitter trigger signal. Standard digital
delay generators have a typical jitter of ≈ 100 ps and even sophisticated electronics does not much
better than ≈ 10 ps. In contrast, the jitter of mode-locked laser pulses is usually in the order of
≈ 1 ps. To benefit from this precision, the trigger pulses are generated by the laser pulses and a
fast photodiode. The two AOMs are driven by a common delay generator (SRG DG 535), which
is synchronized to the laser pulse train. Both AOMs are working at the same pulse repetition
frequency, but with independently selectable delays for on- and off-switching. So, two windows
for laser and trigger are opened, whose jitter does not effect the time interval of the pulses relative
to each other. The temporal distance of the windows is tunable in 12.5 ns-steps (that is the time
between two pulses @ 80 MHz pulse repetition frequency). The jitter between trigger and signal
pulse was measured with a sampling oscilloscope (Tektronix CSA 830) and found to be less than
8 ps, including all electrical noise.
Figure 2a shows a selected laser pulse (lower line) and the transmission window given by the AOM
(upper line). The resolution of this measurement is limited by the bandwidth of the oscilloscope
(1 GHz) and the rise-time of the photodiode (300 ps). The pulse-duration of 8.5 ps is measured with
a background-free intensity autocorrelation, shown in Fig. 2b. Rise- and fall-times as short as 5 ns
are required for the transmission window to provide proper single-pulse switching. Such values
can only be achieved with cutting edge AOMs that generate small Bragg-zones for visible radiation
(e.g. AA.MQ280-vis). The green beam has to be focused to a waist of < 50 m, that corresponds
to a peak power density of > 3.2 GW/cm2 . Precise alignment and a dust-free environment are
premises for the stable operation of this laser system.
3 FHG: Beta-Bariumborat: β-BaB2 O4 (BBO).
AG Magnetismus
90
TU Kaiserslautern
This work is supported by the Stiftung Rheinland-Pfalz für Innovation.
References
[1] LUMERA LASER GmbH, Opelstrasse 10, D−67661 Kaiserslautern, www.LUMERA-LASER.com,
Phone: +496301703180, Fax: +496301703189, Email: info@LUMERA-LASER.com.
AG Magnetismus
91
TU Kaiserslautern
6.18 The influence of amorphous carbon coating on encrustations of in-

dwelling catheter surfaces
L. Kleinen, K. Jung, B. Hillebrands1
The blockage and encrustation of urethral catheters by crystalline bacterial biofilms remains a ma-
jor complication in the care of patients undergoing long-term indwelling bladder catheterization.
Approximately 25 % of all nosocomial infections are related to bladder catheterization. These in-
fections are initiated by microorganisms attaining to the bladder via the catheter. Proliferation of
urease-producing bacteria, e.g. Proteus mirabilis, leads to a splitting of urea into ammonia and
bicarbonate, and subsequently to a dramatic increase in the urinary pH. The alkaline environment
supports the precipitation of phosphate salts which deposit as infection stones on the catheter sur-
face (Fig. 1). Occasionally, the mineral brushite (CaHPO4 · 2H2 O) can be formed at pH levels
between 6.5 and 7.2; at higher pH levels it can transform into carbonate apatite.
In this study we investigated the influence of amorphous carbon coating on surfaces of indwelling
catheters in respect of its ability to reduce the extent of encrustation.
A 50 nm thick carbon layer was deposited on the catheter surfaces by Plasma Enhanced Chemical
Vapour Deposition (PECVD) using a commercially available inductively coupled plasma beam
source (COPRA™ , CCR GmbH Rheinbreitbach). This source is characterized by a very effective
rf excitation of the plasma. The gas pressure can be chosen very low and the delay time of the gas
molecules in the source is short. Therefore predominantly C2 H+ +
2 or C2 H ions leave the plasma
source (Fig. 2). In the deposition process the ion energy and the ion flux are reduced by a factor of
ten with respect to the conditions typical for the production of diamond-like carbon films [1]. Thus
Fig. 1: Reaction scheme of infection stone formation (left). Struvite-carbonate apatite stone which has com-
pletely filled the kidney (right).
1 In
collaboration with H. Busch and U. Grabowy, NTTF GmbH Rheinbreitbach, and N. Laube, Department of
Urology, Universität Bonn.
AG Magnetismus
92
TU Kaiserslautern
Fig. 2: Scheme of the PECVD-

process. Ionized acetylene gas
released by the plasma source is
deposited as amorphous carbon
on the catheter surface.
the energy flux density is reduced by two orders of magnitude. The temperature of the substrate is
practically not increased during the deposition process. Therefore this technique qualifies even for
coating highly temperature sensitive materials (e.g. polymers) with amorphous carbon. Despite
the fact that the impinging ions have only a kinetic energy of 20 eV, the deposited films are rather
hard and scratch resistant with respect to steel. The films exhibit a Vickers hardness of H = 10 GPa
and an elasticity of E = 60 GPa. The ratio E/H of 6 is in agreement with the constraint model of
Angus [2]. That means the carbon films are thermodynamically stable.
For the investigation of the crystallization processes under constant conditions, an in-vitro constant-
flow-crystallizer model and a standardized test procedure were developed (Fig. 3). 24 commercial
indwelling catheters were incubated in synthetic urine for 12h. Urea-splitting was started by the
addition of urease. Encrusted surfaces were analyzed with SEM; the chemical composition and
sample removal urine in-flow Fig. 3: Schematic view of the glas-

(80 ml/h)
silicone stopper bladder-model for standardized inves-
tigation of encrustation processes on
indwelling catheters (top); detailed
view (bottom).
acked „glas bladder“ water out-flow
glas capillar
residual volume
catheter eye
(approx. 25ml urine)
blocked catheter
ater in flow (10 ml H2O)
(37°C)
catheter shaft
silicone stopper
slot
waste urine
pH = 5.6
urine
6.5 gr
adi
ad
en
ie current
t
p dissolution
pH
9.0 Instillagel//urease
mixture
AG Magnetismus
93
TU Kaiserslautern
Fig. 4: Encrusted catheter after

12 h incubation (left); SEM pic-
ture of mineral deposits on un-
coated catheter surface (right).
the amount of the minerals deposited on the catheter surface were determined.
After 12 hours of incubation the catheters (n=24) showed mineral deposits (Fig. 4, left). Encrus-
tation was analyzed with SEM (Fig. 4, right). Crystallized minerals were identified as carbonate
apatite, struvite, and brushite. A 50 % decrease of encrustation was found for coated catheters
(Fig. 5).
Fig. 5: Mean values of

the amounts of stone ma-
terial precipitated within
12 hours of incubation on
the upper 5 cm of the ex-
panded catheters. Dif-
ferent mineral phases are
analyzed: apatite (dark
grey), brushite (grey), stru-
vite (light grey). Left: un-
coated, right: coated.
Therefore carbon coating is a new strategy to improve the surface of indwelling catheters in order to
protect them from rapid encrustation. The results of the presented pilot study clearly show that the
amout of mineral phases which precipitated on the coated test catheters surfaces can be reduced by
50 % compared to untreated catheters [3]. Further experiments will be performed to understand the
mechanisms of this effect, which leads to a broad range of applications. Encrustation reducing thin
films also qualify for coatings on uereter stents and other endwelling instruments for the urinary
tract.
Work supported by the Stiftung Rheinland-Pfalz für Innovation.
References
[1] R. Kleber, M. Weiler, A. Krüger, S. Sattel, G. Kunz, K. Jung, H. Ehrhardt, Diam. Rel. Mat. 2, 246 (1993).
[2] J.C. Angus, P. Koidl, S. Domitz, “Carbon Thin Films”, in: Plasma Deposited Thin Films (Eds.: J. Mort,
F. Jansen), CRC press, Cleveland (1986).
[3] N. Laube, L. Kleinen, K. Schenk, Management und Krankenhaus, August (2003).
AG Magnetismus
94
TU Kaiserslautern
7 Publications
Chapter 7: Publications
Most publications can be downloaded from http://www.physik.uni-kl.de/hillebrands.
7.1 published
1. Confined dynamicc excitations in structured magnetic media
S.O. Demokritov, B. Hillebrands, A.N. Slavin
IEEE Tran. Mag. 38, 2502 (2002).
2. Anisotropy of magneto-optical spectra in ultrathin Fe/Au/Fe bilayers

J. Grondilova, M. Rickart, J. Mistrik, K. Postava, S. Visnovsky, T. Yamaguchi, R. Lopusnik,
S.O. Demokritov, B. Hillebrands
J. Appl. Phys. 91, 8246 (2002).
3. Ultrafast magnetic switching

B. Hillebrands, J. Fassbender
Nature 418, 493 (2002).
4. Effective dipolar boundary conditions for dynamic magnetization in thin magnetic stripes
K. Yu. Guslienko, S.O. Demokritov, B. Hillebrands, A.N. Slavin
Phys. Rev. B 66, 132402 (2002).
5. Interlayer interaction in a Fe/Cr/Fe System: Dependence on the thickness of the chrome

interlayer and on temperature
S.O. Demokritov, A.B. Drovosekov, N.M. Kreines, H. Nembach, M. Rickart, D.J. Kholin
Journal of Experimental and Theoretical Physics 95, 1062 (2002).
6. Temperature dependence of interlayer coupling in Fe/Cr/Fe layered system

S. O. Demokritov, A.B. Drovosekov, N.M. Kreines, H. Nembach, M. Rickart, D.I. Kholin,
Sov. Phys. JETP 95, 10621 (2002).
7. Spin dynamics in confined magnetic structures II

B. Hillebrands, K. Ounadjela (Eds)
Springer, Berlin (2003).
8. Magnetization dynamics investigated by time-resolved Kerr effect magnetometry

J. Fassbender
in: Spin dynamics in confined magnetic structures II, Eds. B. Hillebrands, K. Ounadjela
Springer, Berlin (2003).
9. Confined dynamic excitations in structured magnetic media

S.O. Demokritov, B. Hillebrands, A.N. Slavin
IEEE Trans. Magn. 38, 2502 (2003).
10. Temperature dependence of interlayer coupling in a Fe/Cr/Fe wedge sample, MOKE and
MBLS studies
S. O. Demokritov, A.B. Drovosekov, D.I. Kholin, N.M. Kreines
J. Magn. Magn. Mater. 258-259, 391 (2003).
AG Magnetismus
95
TU Kaiserslautern
7 Publications
11. Collison properties of quasi-one-dimensional spin wave solitons and two-dimensional spin
wave bullets
A.N. Slavin, O. Büttner, M. Bauer, S.O. Demokritov, B. Hillebrands, M.P. Kostylev,
B.A. Kalinikos, V.V. Grimalsky, Yu. Rapoport
Chaos, 13, 693 (2003).
12. Coherent precessional magnetization reversal in microscopic magnetic memory elements

H.W. Schumacher, C. Chappert, P. Crozat, R.C. Sousa, P.P. Freitas, J. Miltat, J. Fassbender,
B. Hillebrands
Phys. Rev. Lett. 90, 017201 (2003).
13. Control of interlayer exchange coupling in Fe/Cr/Fe trilayers by ion beam irradiation
S.O. Demokritov, C. Bayer, S. Poppe, M. Rickart, J. Fassbender, B. Hillebrands, D.I. Kholin,
N.M. Kreines, O.M. Liedke
Phys. Rev. Lett. 90, 097201 (2003).
14. Probing interface magnetism in the FeMn/NiFe exchange bias system using magnetic second-
harmonic generation
L.C. Sampaio, A. Mougin, J. Ferré, P. Georges, A. Brun, H. Bernas, S. Poppe, T. Mewes,
J. Fassbender, B. Hillebrands
Europhys. Lett. 63, 819 (2003).
15. Ion irradiation of exchange-bias systems for magnetic sensor applications

J. Fassbender, S. Poppe, T. Mewes, J. Jurszek, B. Hillebrands, D. Engel, M. Jung, A. Ehres-
mann, H. Schmoranzer, K.U. Barholz, R. Mattheis
Appl. Phys. A 77, 51 (2003).
16. Phase diagrams and energy barriers of exchange-biased bilayers with additional anisotropies
in the ferromagnet
T. Mewes, H. Nembach, J. Fassbender, B. Hillebrands, J.V. Kim, R.L. Stamps
Phys. Rev. B 67, 104422 (2003).
17. Formation of envelope solitons from parametrically amplified and conjugated spin wave
pulses
A.A. Serga, S.O. Demokritov, B. Hillebrands, A.N. Slavin
J. Appl. Phys. 93, 8758 (2003).
18. Phase control of non-adiabatic parametric amplification of spin wave packets

A.A. Serga, S.O. Demokritov, B. Hillebrands, Seong-Gi Min, A.N. Slavin
J. Appl. Phys. 93, 8585 (2003).
19. Spin wave wells with multiple states created in small magnetic elements
C. Bayer, S.O. Demokritov, B. Hillebrands, A.N. Slavin
Appl. Phys. Lett. 82, 607 (2003).
20. Dynamics in magnetic stripes

B. Hillebrands
Festschrift in honor of Prof. Dr. Gernot Güntherodts 60th anniversary, Aachen, May 2 (2003).
AG Magnetismus
96
TU Kaiserslautern
7 Publications
21. Ultraschnelle magnetische Speicher

J. Fassbender, B. Hillebrands
Physik in unserer Zeit 34, 102 (2003).
22. Parametric generation of soliton-like spinwave packets in active rings based on ferromag-
netic films
A.A. Serga, M. Kostylev, B.A. Kalinikos, S.O. Demokritov, B. Hillebrands, H. Benner
JETP Lett. 77, 350 (2003).
23. Amorphe Kohlenstoffschichten, Geeignete Materialien zur Beeinflussung von Katheterenkrus-
tationen
N. Laube, L. Kleinen, K. Schenk
Management und Krankenhaus, August (2003).
7.2 in press
1. Brillouin light scattering spectroscopy
B. Hillebrands
in: Novel techniques for characterizing magnetic materials, Yimey Zhu, editor, Kluwer Aca-
demic Press, in press.
2. Induced four fold anisotropy and bias in compensated NiFe/FeMn double layers
T. Mewes, B. Hillebrands, R.L. Stamps
Phys. Rev. B, in press.
3. Experimental observation of symmetry breaking nonlinear modes in an active ring
S.O. Demokritov, A.A. Serga, V.E. Demidov, B. Hillebrands, M. Kostylev, B.A. Kalinikos
Nature 426, 159 (2003), in press.
4. Magnetismus
B. Hillebrands, S. Blügel
in: Bergmann-Schäfer, Experimentalphysik Band 6 (Festkörper)
de Gruyters Verlag, in press.
7.3 submitted
1. On the mechanism of irradiation enhanced exchange bias
S. Poppe, J. Fassbender, B. Hillebrands
submitted to Appl. Phys. Lett..
2. General methods for the determination of the stiffness tensor and mass density of thin films
using Brillouin light scattering: study on tungsten carbide films
T. Wittkowski, G. Distler, K. Jung, B. Hillebrands, J.D. Comins
submitted to Phys. Rev. B..
3. Low energetic ion embedding by ECWR plasma, the given production technology to form
TMR barriers for MRAM
W. Maas, J. Langer, B. Ocker, S.O. Demokritov, B. Hillebrands, B.F.P. Roos, M. Weiler
submitted to Vacuum.
AG Magnetismus
97
TU Kaiserslautern
7 Publications
4. Magnetic interlayer coupling across semiconducting EuS layers

U. Rücker, S.O. Demokritov, J. Nassar, P. Grünberg
submitted to Europhys. Lett..
5. Self-Generation of Two-Dimensional Spin-Wave Bullets

A.A. Serga, S.O. Demokritov, B. Hillebrands, A.N. Slavin
submitted to Phys. Rev. Lett..
6. Separation of the three magnetization vector components for time-resolved magneto-optic

Kerr effect measurements
H. Nembach, M.C. Weber, J. Fassbender, B. Hillebrands
submitted to J. Appl. Phys..
7. Picosecond optical control of the magnetization in exchange biased NiFe/FeMn bilayers

M.C. Weber, H. Nembach, J. Fassbender
submitted to J. Appl. Phys..
8. High resolution magnetic patterning of exchange coupled multilayers

V.E. Demidov, D.I. Kholin, S.O. Demokritov, B. Hillebrands
submitted to Appl. Phys. Lett..
7.4 Ph.D. theses

1. Systematik epitaktischer magnetischer Austausch-Verschiebungs-Schichtsysteme
Tim Mewes, Ph.D. thesis, Universität Kaiserslautern, Nov. 2002.
2. Modifikation des Austausch-Verschiebungseffektes in Ferro-/Antiferromagnet-Bilagen durch

lagenweise und ortsaufgelöste Ionenbestrahlung
Stefan Poppe, Ph.D. thesis, Universität Kaiserslautern, May 2002.
AG Magnetismus
98
TU Kaiserslautern
8 Conferences, Workshops, Schools and Seminars
Chapter 8: Conferences, Workshops, Schools and Seminars

(shown in chronological order with the speaker named)
8.1 Conferences
8.1.1 Invited talks
S.O. Demokritov:
Spin wave dynamics in structured magnetic media
Congress American Vacuum Society, Denver, USA, November 2002
B. Hillebrands:
Dynamics in patterned magnetic structures
2002 MRS-Fall-Meeting, Boston, USA, December 2002
J. Fassbender:
Maßgeschneiderte Materialien für die Magnetoelektronik
Symposium “Grundlagenorientierte Materialwissenschaften”,
Universität Braunschweig, Germany, December 2002
B. Hillebrands:
Dynamik in magnetischen Nanostrukturen
DFG-Rundgespräch, Bad Honnef, Germany, March 2003
B. Hillebrands:
Dynamics in magnetic stripes
LEA workshop, Grenoble, France, May 2003
B. Hillebrands:
Fast magnetization dynamics
ICM 2003, Rom, July/August 2003
8.1.2 Contributed talks and posters
5 contribution: 47th Annual Conference on Magnetism and Magnetic Materials

Tampa, Florida, USA, November 2002
10 contributions: DPG-Frühjahrstagung Dresden, Germany, March 2003
1 contribution: ICMCTF, San Diego, USA, April/May 2003
1 contribution: E-MRS Tagung, Strasbourg, France, June 2003
4 contributions: 18th ICMFS Conference, Madrid, July 2003
3 contributions: ICM 2003, Rom, July/August 2003
1 contribution: Diamond 2003, Salzburg, Austria, September 2003
1 contribution: FKA 12, Wien, Austria, September 2003
1 contribution: ECASIA 2003, Berlin, Germany, October 2003
1 contribution: 18th European Conference on Biomaterials, Stuttgart, Germany, October 2003
AG Magnetismus
99
TU Kaiserslautern
8.2 Invited colloquia and seminars

S.O. Demokritov:
Dynamics in patterned magnetic stuctures
IBM, New York, USA, November 2002
B. Hillebrands:
Spin waves: a model system to study nonlinear wave propagation and amplification in confined
anisotropic materials
CREOLE, University of Central Florida, Orlando, USA, November 2002
J. Fassbender:
Magnetisierungsdynamik und Schaltverhalten von dünnen magnetischen Schichten
FU Berlin, Festkörperphysikalische Kolloquium, December 2002
B. Hillebrands:
Dynamik in magnetischen Mikro- und Nanostrukturen
Universität Münster, Physikalische Kolloquium, Germany, January 2003
S.O. Demokritov:
Spin wave dynamics in structured magnetic media: closer to reality
University of Nebraska, Lincoln, USA, February 2003
S.O. Demokritov:
Spin wave dynamics in strongly non-uniform magnetic fields
University of Ferrara, Italy, March 2003
T. Wittkowski:
Brillouin-Lichtstreuung zur Charakterisierung elastischer und struktureller Eigenschaften
dünner Schichten
Universität Chemnitz, Germany, March 2003
T. Wittkowski:
Untersuchung dünner Schichten mittels Brillouin-Streuung
Fachhochschule Mittweida, Germany, March 2003
T. Wittkowski:
Brillouin-Lichtstreuung zur Charakterisierung elastischer und struktureller Eigenschaften
dünner Schichten
Institut für Physik, TU Chemnitz, Germany, May 2003
C. Bayer:
Spin wave dynamics in structured magnetic media
University of Minnesota, MRSEC-Seminar, USA, May 2003
B. Hillebrands:
Dynamics in magnetic stripes
Festkolloquium anlässlich des 60. Geburtstags von Prof. Dr. Gernot Güntherodt, RWTH
Aachen, Germany, May 2003
AG Magnetismus
100
TU Kaiserslautern
B. Hillebrands:
Dynamik in magnetischen Mikro- und Nanostrukturen
Universität Göttingen, Physikalisches Kolloquium, Germany, May 2003
J. Fassbender:
Funktionale magnetische Schichten - neue Möglichkeiten durch Ionenbestrahlung
Forschungszentrum Rossendorf, Dresden, Germany, August 2003
8.3 Invited lectures

S.O. Demokritov:
Magnetic dynamics in structured systems
University of Perugia, Italy, March 2003
J. Fassbender:
Basics of magnetization dynamics
Training days of the EU-RTN network ULTRASWITCH, Leuven, Oct. 9 − 10, 2003
8.4 Contributions to other meetings

S. Poppe:
Maßgeschneiderte magnetische Materialien
3. Ionenstrahlmeeting, Stuttgart, Germany, November 2002
J. Fassbender:
Konzepte zum Schalten der Magnetisierung - ein Weg zu schnelleren magnetischen Speichern
DFG-Graduierten-Kolleg ”Nichtlineare Optik und Ultrakurzzeitphysik”, Kaiserslautern, Ger-
many, June 2003
J. Fassbender:
Magnetische Strukturierung mittels Ionenstrahlen
Universität Augsburg, Germany, June 2003
J. Fassbender:
Dünne magnetische Schichten
IFOS Kaiserslautern, Germany, July 2003
J. Fassbender:
Magnetische Strukturierung mittels Ionenstrahlen
Ionenstrahlmeeting Dresden, September 2003
T. Wittkowski:
Brillouin-Lichtsteuung zum Nachweis akustischer Oberflächenphononen in BN-beschichteten
dünnen Fasern
10th c-BN Expert Meeting, Mittweida, Germany, October 2003
AG Magnetismus
101
TU Kaiserslautern
Appendix
Appendix: Impressions from our Group Ceremonies
AG Magnetismus
103
TU Kaiserslautern

Dynamic Magnetic Phenomena - Ar2003 PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Dynamic Magnetic Phenomena - Ar2003 PDF

Hochgeladen von

Copyright:

Verfügbare Formate

Annual Report 2003

Postal address: Postfach 3049

From left to right:

This report contains unpublished results and should

D. Exchange Bias Effect .......................................................... 62

Dear colleagues and friends,

Kaiserslautern, November 2003

2.1 Members of the group

Engineers and Technicians

2.2 Visiting scientists, postdoctoral fellows and exchange students

Prof. Robert L. Stamps, Department of Physics,

Prof. J. Darrell Comins, Department of Physics,

Prof. Andrei Slavin, University of Rochester, Michigan, U.S.A. 04.06.03 - 22.06.03

Dimitry Kholin, Ph.D. student, Institute for Physical Problems,

Dr. Mikhail Kostylev, Electronics Faculty,

Markus Laufenberg, Ph.D. student,

2.3 Guest seminars

Prof. Robert D. Shull NIST, Gaithersburg, MD, USA

Dr. Claude Chappert CNRS, Université Paris-Sud, Orsay, France

Prof. J. Darrell Comins Department of Physics, University of the Witwatersrand,

Dr. Radek Lopusnik NIST, Boulder, Colorado, USA

Dr. Britta Hausmanns Universität Duisburg-Essen

Prof. Robert L. Stamps University of Western Australia, Australia

2.4 Visits of group members at other laboratories

Christian Bayer University of Minnesota, USA

Steffen Blomeier University of Glasgow, United Kingdom

Alexander André Oakland University, Rochester, USA

Sergej O. Demokritov University Perugia, Italy

2.5 Group member photo gallery

Alexander André Christian Bayer Andreas Beck

Dr. Mireia Blanco-Mantecón Steffen Blomeier Dr. Vladislav Demidov

Dirk Hoffmann Dr. Kurt Jung Lisa Kleinen

Dr. Galina Khlyap Maciej Oskar Liedke Patricia Martin Pimentel

Dr. Tim Mewes Günther Moog Sibylle Müller

Hans Nembach Bernd Pfaff Dr. Stefan Poppe

Dr. Marc Rickart Martin Rohmer Dr. Björn Roos

Heike Schuster Dr. Alexander Serha Markus Weber

Dieter Weller Dr. Thomas Wittkowski Dr. Q.F. Xiao

Chapter 3: Research Topics

3) Dynamic magnetic properties of laterally patterned nanostructures

4) Nonlinear properties of high-amplitude spin waves

5) Fast magnetic switching

7) Exchange bias systems

8) Elastic properties of hard, super-hard and inhomogeneous films and multilayers

9) Biofunctionalized surfaces for medical applications

8. magnetic force microscope with magnet (NT-MDT)

C) Equipment at the Institute for Thin Film Technologies (IDST), Rheinbreitach

1. Preparation of thin films:

k. Rutherford Backscattering (RBS)2 : analysis of trace elements with depth resolu-

Chapter 5: Transfer of Technology

• magnetic anisotropies, optionally with high spatial resolution

• test of homogeneity of magnetic parameters

• exchange stiffness constants in magnetic films

2. Institut für Dünnschichttechnologie (IDST) - Transferstelle der Tech-

• development of product specific coatings

• optimization of coatings especially for medical applications

• coating of polymers and temperature sensitive materials

• coating of samples and small scale production series

• management for R&D-projects

Please contact us for more information.

Chapter 6: Experimental Results

A. Dynamic Magnetic Phenomena

Fig. 2: Schematic spectrum